GCC Code Coverage Report

Directory:	./
File:	strings/ctype-uca.cc
Date:	2022-12-06 21:40:42

	Exec	Total	Coverage
Lines:	1622	1782	91.0%
Branches:	2616	9044	28.9%

1

2

3

This library is free software; you can redistribute it and/or

4

modify it under the terms of the GNU Library General Public

5

License, version 2.0, as published by the Free Software Foundation.

6

7

This library is also distributed with certain software (including

8

but not limited to OpenSSL) that is licensed under separate terms,

9

as designated in a particular file or component or in included license

10

documentation. The authors of MySQL hereby grant you an additional

11

permission to link the library and your derivative works with the

12

separately licensed software that they have included with MySQL.

13

14

Without limiting anything contained in the foregoing, this file,

15

which is part of C Driver for MySQL (Connector/C), is also subject to the

16

Universal FOSS Exception, version 1.0, a copy of which can be found at

17

http://oss.oracle.com/licenses/universal-foss-exception.

18

19

This library is distributed in the hope that it will be useful,

20

but WITHOUT ANY WARRANTY; without even the implied warranty of

21

MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU

22

Library General Public License, version 2.0, for more details.

23

24

You should have received a copy of the GNU Library General Public

25

License along with this library; if not, write to the Free

26

Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston,

MA 02110-1301 USA */

/*

UCA (Unicode Collation Algorithm) support.

31

32

Features that are not implemented yet:

33

- No Normalization From D is done

34

+ No decomposition is done

35

+ No Thai/Lao orderding is done

36

- No combining marks processing is done

*/

#include <assert.h>

#include <stdio.h>

#include <string.h>

#include <sys/types.h>

#include <algorithm>

#include <bitset>

#include <iterator>

#include <map>

#include <utility>

#include "m_ctype.h"

#include "m_string.h"

51

#include "my_byteorder.h"

52

#include "my_compiler.h"

53

54

#include "my_inttypes.h"

55

#include "my_loglevel.h"

56

#include "my_macros.h"

57

#include "mysys_err.h"

58

#include "strings/mb_wc.h"

59

#include "strings/str_uca_type.h"

60

#include "strings/uca900_data.h"

61

#include "strings/uca900_ja_data.h"

62

#include "strings/uca900_zh_data.h"

63

#include "strings/uca_data.h"

64

#include "template_utils.h"

65

66

MY_UCA_INFO my_uca_v400 = {

67

UCA_V400,

68

69

0xFFFF, /* maxchar */

70

uca_length, uca_weight, false, nullptr, /* contractions */

71

nullptr,

72

73

/* Logical positions */

74

0x0009, /* first_non_ignorable p != ignore */

75

0xA48C, /* last_non_ignorable Not a CJK and not UNASSIGNED */

76

77

0x0332, /* first_primary_ignorable p == 0 */

78

0x20EA, /* last_primary_ignorable */

79

80

0x0000, /* first_secondary_ignorable p,s == 0 */

81

0xFE73, /* last_secondary_ignorable p,s == 0 */

82

83

0x0000, /* first_tertiary_ignorable p,s,t == 0 */

84

0xFE73, /* last_tertiary_ignorable p,s,t == 0 */

85

86

0x0000, /* first_trailing */

87

0x0000, /* last_trailing */

88

89

0x0009, /* first_variable */

90

0x2183, /* last_variable */

91

0, /* extra_ce_pri_base, not used */

92

0, /* extra_ce_sec_base, not used */

93

0 /* extra_ce_ter_base, not used */

94

};

95

96

/******************************************************/

97

98

MY_UCA_INFO my_uca_v520 = {

99

UCA_V520,

100

101

0x10FFFF, /* maxchar */

uca520_length,

uca520_weight,

false,

nullptr, /* contractions */

106

nullptr,

107

108

0x0009, /* first_non_ignorable p != ignore */

109

0x1342E, /* last_non_ignorable Not a CJK and not UASSIGNED */

110

111

0x0332, /* first_primary_ignorable p == ignore */

112

0x101FD, /* last_primary_ignorable */

113

114

0x0000, /* first_secondary_ignorable p,s= ignore */

115

0xFE73, /* last_secondary_ignorable */

116

117

0x0000, /* first_tertiary_ignorable p,s,t == ignore */

118

0xFE73, /* last_tertiary_ignorable */

119

120

0x0000, /* first_trailing */

121

0x0000, /* last_trailing */

122

123

0x0009, /* first_variable if alt=non-ignorable: p != ignore */

124

0x1D371, /* last_variable if alt=shifter: p,s,t == ignore */

125

0, /* extra_ce_pri_base, not used */

126

0, /* extra_ce_sec_base, not used */

127

0 /* extra_ce_ter_base, not used */

128

};

129

130

/******************************************************/

/*

German Phonebook

*/

static const char german2[] =

136

"&AE << \\u00E6 <<< \\u00C6 << \\u00E4 <<< \\u00C4 "

137

"&OE << \\u0153 <<< \\u0152 << \\u00F6 <<< \\u00D6 "

138

"&UE << \\u00FC <<< \\u00DC ";

139

140

/*

141

Some sources treat LETTER A WITH DIAERESIS (00E4,00C4)

142

secondary greater than LETTER AE (00E6,00C6).

143

http://www.evertype.com/alphabets/icelandic.pdf

144

http://developer.mimer.com/collations/charts/icelandic.htm

145

146

Other sources do not provide any special rules

147

for LETTER A WITH DIAERESIS:

148

http://www.omniglot.com/writing/icelandic.htm

149

http://en.wikipedia.org/wiki/Icelandic_alphabet

150

http://oss.software.ibm.com/icu/charts/collation/is.html

151

152

Let's go the first way.

153

*/

154

155

static const char icelandic[] =

156

"& A < \\u00E1 <<< \\u00C1 "

157

"& D < \\u00F0 <<< \\u00D0 "

158

"& E < \\u00E9 <<< \\u00C9 "

159

"& I < \\u00ED <<< \\u00CD "

160

"& O < \\u00F3 <<< \\u00D3 "

161

"& U < \\u00FA <<< \\u00DA "

162

"& Y < \\u00FD <<< \\u00DD "

163

"& Z < \\u00FE <<< \\u00DE "

164

"< \\u00E6 <<< \\u00C6 << \\u00E4 <<< \\u00C4 "

165

"< \\u00F6 <<< \\u00D6 << \\u00F8 <<< \\u00D8 "

166

"< \\u00E5 <<< \\u00C5 ";

167

168

/*

169

Some sources treat I and Y primary different.

170

Other sources treat I and Y the same on primary level.

171

We'll go the first way.

172

*/

173

174

static const char latvian[] =

175

"& C < \\u010D <<< \\u010C "

176

"& G < \\u0123 <<< \\u0122 "

177

"& I < \\u0079 <<< \\u0059 "

178

"& K < \\u0137 <<< \\u0136 "

179

"& L < \\u013C <<< \\u013B "

180

"& N < \\u0146 <<< \\u0145 "

181

"& R < \\u0157 <<< \\u0156 "

182

"& S < \\u0161 <<< \\u0160 "

183

"& Z < \\u017E <<< \\u017D ";

184

185

static const char romanian[] =

186

"& A < \\u0103 <<< \\u0102 < \\u00E2 <<< \\u00C2 "

187

"& I < \\u00EE <<< \\u00CE "

188

"& S < \\u0219 <<< \\u0218 << \\u015F <<< \\u015E "

189

"& T < \\u021B <<< \\u021A << \\u0163 <<< \\u0162 ";

190

191

static const char slovenian[] =

192

"& C < \\u010D <<< \\u010C "

193

"& S < \\u0161 <<< \\u0160 "

194

"& Z < \\u017E <<< \\u017D ";

195

196

static const char polish[] =

197

"& A < \\u0105 <<< \\u0104 "

198

"& C < \\u0107 <<< \\u0106 "

199

"& E < \\u0119 <<< \\u0118 "

200

"& L < \\u0142 <<< \\u0141 "

201

"& N < \\u0144 <<< \\u0143 "

202

"& O < \\u00F3 <<< \\u00D3 "

203

"& S < \\u015B <<< \\u015A "

204

"& Z < \\u017A <<< \\u0179 < \\u017C <<< \\u017B";

205

206

static const char estonian[] =

207

"& S < \\u0161 <<< \\u0160 "

208

" < \\u007A <<< \\u005A "

209

" < \\u017E <<< \\u017D "

210

"& W < \\u00F5 <<< \\u00D5 "

211

"< \\u00E4 <<< \\u00C4 "

212

"< \\u00F6 <<< \\u00D6 "

213

"< \\u00FC <<< \\u00DC ";

214

215

// Standard Spanish, also for Galician.

216

static const char spanish[] = "& N < \\u00F1 <<< \\u00D1 ";

217

218

/*

219

Some sources treat V and W as similar on primary level.

220

We'll treat V and W as different on primary level.

221

*/

222

223

static const char swedish[] =

224

"& Y <<\\u00FC <<< \\u00DC "

225

"& Z < \\u00E5 <<< \\u00C5 "

226

"< \\u00E4 <<< \\u00C4 << \\u00E6 <<< \\u00C6 "

227

"< \\u00F6 <<< \\u00D6 << \\u00F8 <<< \\u00D8 ";

228

229

static const char turkish[] =

230

"& C < \\u00E7 <<< \\u00C7 "

231

"& G < \\u011F <<< \\u011E "

232

"& H < \\u0131 <<< \\u0049 "

233

"& O < \\u00F6 <<< \\u00D6 "

234

"& S < \\u015F <<< \\u015E "

235

"& U < \\u00FC <<< \\u00DC ";

236

237

static const char czech[] =

238

"& C < \\u010D <<< \\u010C "

239

"& H < ch <<< Ch <<< CH"

240

"& R < \\u0159 <<< \\u0158"

241

"& S < \\u0161 <<< \\u0160"

242

"& Z < \\u017E <<< \\u017D";

243

244

static const char danish[] = /* Also good for Norwegian */

245

"& Y << \\u00FC <<< \\u00DC << \\u0171 <<< \\u0170"

246

"& Z < \\u00E6 <<< \\u00C6 << \\u00E4 <<< \\u00C4"

247

" < \\u00F8 <<< \\u00D8 << \\u00F6 <<< \\u00D6 << \\u0151 <<< \\u0150"

248

" < \\u00E5 <<< \\u00C5 << aa <<< Aa <<< AA";

249

250

static const char lithuanian[] =

251

"& C << ch <<< Ch <<< CH< \\u010D <<< \\u010C"

252

"& E << \\u0119 <<< \\u0118 << \\u0117 <<< \\u0116"

253

"& I << y <<< Y"

254

"& S < \\u0161 <<< \\u0160"

255

"& Z < \\u017E <<< \\u017D";

256

257

static const char slovak[] =

258

"& A < \\u00E4 <<< \\u00C4"

259

"& C < \\u010D <<< \\u010C"

260

"& H < ch <<< Ch <<< CH"

261

"& O < \\u00F4 <<< \\u00D4"

262

"& S < \\u0161 <<< \\u0160"

263

"& Z < \\u017E <<< \\u017D";

264

265

static const char spanish2[] = /* Also good for Asturian and Galician */

266

"&C < ch <<< Ch <<< CH"

267

"&L < ll <<< Ll <<< LL"

268

"&N < \\u00F1 <<< \\u00D1";

269

270

static const char roman[] = /* i.e. Classical Latin */

"& I << j <<< J "

"& V << u <<< U ";

/*

Persian collation support was provided by

276

Jody McIntyre <mysql@modernduck.com>

277

278

To: internals@lists.mysql.com

279

Subject: Persian UTF8 collation support

280

Date: 17.08.2004

281

282

Contraction is not implemented. Some implementations do perform

283

contraction but others do not, and it is able to sort all my test

strings correctly.

Jody.

*/

static const char persian[] =

289

"& \\u066D < \\u064E < \\uFE76 < \\uFE77 < \\u0650 < \\uFE7A < \\uFE7B"

290

" < \\u064F < \\uFE78 < \\uFE79 < \\u064B < \\uFE70 < \\uFE71"

291

" < \\u064D < \\uFE74 < \\u064C < \\uFE72"

292

"& \\uFE7F < \\u0653 < \\u0654 < \\u0655 < \\u0670"

293

"& \\u0669 < \\u0622 < \\u0627 < \\u0671 < \\u0621 < \\u0623 < \\u0625"

294

" < \\u0624 < \\u0626"

295

"& \\u0642 < \\u06A9 < \\u0643"

296

"& \\u0648 < \\u0647 < \\u0629 < \\u06C0 < \\u06CC < \\u0649 < \\u064A"

297

"& \\uFE80 < \\uFE81 < \\uFE82 < \\uFE8D < \\uFE8E < \\uFB50 < \\uFB51"

298

" < \\uFE80 "

299

/*

300

FE80 appears both in reset and shift.

301

We need to break the rule here and reset to *new* FE80 again,

302

so weight for FE83 is calculated as P[FE80]+1, not as P[FE80]+8.

303

*/

304

" & \\uFE80 < \\uFE83 < \\uFE84 < \\uFE87 < \\uFE88 < \\uFE85"

305

" < \\uFE86 < \\u0689 < \\u068A"

306

"& \\uFEAE < \\uFDFC"

307

"& \\uFED8 < \\uFB8E < \\uFB8F < \\uFB90 < \\uFB91 < \\uFED9 < \\uFEDA"

308

" < \\uFEDB < \\uFEDC"

309

"& \\uFEEE < \\uFEE9 < \\uFEEA < \\uFEEB < \\uFEEC < \\uFE93 < \\uFE94"

310

" < \\uFBA4 < \\uFBA5 < \\uFBFC < \\uFBFD < \\uFBFE < \\uFBFF"

311

" < \\uFEEF < \\uFEF0 < \\uFEF1 < \\uFEF2 < \\uFEF3 < \\uFEF4"

312

" < \\uFEF5 < \\uFEF6 < \\uFEF7 < \\uFEF8 < \\uFEF9 < \\uFEFA"

313

" < \\uFEFB < \\uFEFC";

/*

Esperanto tailoring.

Contributed by Bertilo Wennergren <bertilow at gmail dot com>

318

September 1, 2005

319

*/

320

static const char esperanto[] =

321

"& C < \\u0109 <<< \\u0108"

322

"& G < \\u011D <<< \\u011C"

323

"& H < \\u0125 <<< \\u0124"

324

"& J < \\u0135 <<< \\u0134"

325

"& S < \\u015d <<< \\u015c"

326

"& U < \\u016d <<< \\u016c";

327

328

/*

329

A simplified version of Hungarian, without consonant contractions.

330

*/

331

static const char hungarian[] =

332

"&O < \\u00F6 <<< \\u00D6 << \\u0151 <<< \\u0150"

333

"&U < \\u00FC <<< \\u00DC << \\u0171 <<< \\u0170";

334

335

static const char croatian[] =

336

"&C < \\u010D <<< \\u010C < \\u0107 <<< \\u0106"

337

"&D < d\\u017E = \\u01C6 <<< d\\u017D <<< D\\u017E = \\u01C5 <<< D\\u017D "

338

"= \\u01C4"

339

" < \\u0111 <<< \\u0110"

340

"&L < lj = \\u01C9 <<< lJ <<< Lj = \\u01C8 <<< LJ = \\u01C7"

341

"&N < nj = \\u01CC <<< nJ <<< Nj = \\u01CB <<< NJ = \\u01CA"

342

"&S < \\u0161 <<< \\u0160"

343

"&Z < \\u017E <<< \\u017D";

344

345

/*

346

SCCII Part 1 : Collation Sequence (SLS1134)

347

2006/11/24

348

Harshula Jayasuriya <harshula at gmail dot com>

349

Language Technology Research Lab, University of Colombo / ICTA

350

*/

351

#if 0

352

static const char sinhala[]=

353

"& \\u0D96 < \\u0D82 < \\u0D83"

354

"& \\u0DA5 < \\u0DA4"

355

"& \\u0DD8 < \\u0DF2 < \\u0DDF < \\u0DF3"

356

"& \\u0DDE < \\u0DCA";

357

#else

358

static const char sinhala[] =

359

"& \\u0D96 < \\u0D82 < \\u0D83 < \\u0D9A < \\u0D9B < \\u0D9C < \\u0D9D"

360

"< \\u0D9E < \\u0D9F < \\u0DA0 < \\u0DA1 < \\u0DA2 < \\u0DA3"

361

"< \\u0DA5 < \\u0DA4 < \\u0DA6"

362

"< \\u0DA7 < \\u0DA8 < \\u0DA9 < \\u0DAA < \\u0DAB < \\u0DAC"

363

"< \\u0DAD < \\u0DAE < \\u0DAF < \\u0DB0 < \\u0DB1"

364

"< \\u0DB3 < \\u0DB4 < \\u0DB5 < \\u0DB6 < \\u0DB7 < \\u0DB8"

365

"< \\u0DB9 < \\u0DBA < \\u0DBB < \\u0DBD < \\u0DC0 < \\u0DC1"

366

"< \\u0DC2 < \\u0DC3 < \\u0DC4 < \\u0DC5 < \\u0DC6"

367

"< \\u0DCF"

368

"< \\u0DD0 < \\u0DD1 < \\u0DD2 < \\u0DD3 < \\u0DD4 < \\u0DD6"

369

"< \\u0DD8 < \\u0DF2 < \\u0DDF < \\u0DF3 < \\u0DD9 < \\u0DDA"

370

"< \\u0DDB < \\u0DDC < \\u0DDD < \\u0DDE < \\u0DCA";

371

#endif

372

373

static const char vietnamese[] =

374

" &A << \\u00E0 <<< \\u00C0" /* A */

375

" << \\u1EA3 <<< \\u1EA2"

376

" << \\u00E3 <<< \\u00C3"

377

" << \\u00E1 <<< \\u00C1"

378

" << \\u1EA1 <<< \\u1EA0"

379

" < \\u0103 <<< \\u0102" /* A WITH BREVE */

380

" << \\u1EB1 <<< \\u1EB0"

381

" << \\u1EB3 <<< \\u1EB2"

382

" << \\u1EB5 <<< \\u1EB4"

383

" << \\u1EAF <<< \\u1EAE"

384

" << \\u1EB7 <<< \\u1EB6"

385

" < \\u00E2 <<< \\u00C2" /* A WITH CIRCUMFLEX */

386

" << \\u1EA7 <<< \\u1EA6"

387

" << \\u1EA9 <<< \\u1EA8"

388

" << \\u1EAB <<< \\u1EAA"

389

" << \\u1EA5 <<< \\u1EA4"

390

" << \\u1EAD <<< \\u1EAC"

391

" &D < \\u0111 <<< \\u0110" /* D WITH STROKE */

392

" &E << \\u00E8 <<< \\u00C8" /* E */

393

" << \\u1EBB <<< \\u1EBA"

394

" << \\u1EBD <<< \\u1EBC"

395

" << \\u00E9 <<< \\u00C9"

396

" << \\u1EB9 <<< \\u1EB8"

397

" < \\u00EA <<< \\u00CA" /* E WITH CIRCUMFLEX */

398

" << \\u1EC1 <<< \\u1EC0"

399

" << \\u1EC3 <<< \\u1EC2"

400

" << \\u1EC5 <<< \\u1EC4"

401

" << \\u1EBF <<< \\u1EBE"

402

" << \\u1EC7 <<< \\u1EC6"

403

" &I << \\u00EC <<< \\u00CC" /* I */

404

" << \\u1EC9 <<< \\u1EC8"

405

" << \\u0129 <<< \\u0128"

406

" << \\u00ED <<< \\u00CD"

407

" << \\u1ECB <<< \\u1ECA"

408

" &O << \\u00F2 <<< \\u00D2" /* O */

409

" << \\u1ECF <<< \\u1ECE"

410

" << \\u00F5 <<< \\u00D5"

411

" << \\u00F3 <<< \\u00D3"

412

" << \\u1ECD <<< \\u1ECC"

413

" < \\u00F4 <<< \\u00D4" /* O WITH CIRCUMFLEX */

414

" << \\u1ED3 <<< \\u1ED2"

415

" << \\u1ED5 <<< \\u1ED4"

416

" << \\u1ED7 <<< \\u1ED6"

417

" << \\u1ED1 <<< \\u1ED0"

418

" << \\u1ED9 <<< \\u1ED8"

419

" < \\u01A1 <<< \\u01A0" /* O WITH HORN */

420

" << \\u1EDD <<< \\u1EDC"

421

" << \\u1EDF <<< \\u1EDE"

422

" << \\u1EE1 <<< \\u1EE0"

423

" << \\u1EDB <<< \\u1EDA"

424

" << \\u1EE3 <<< \\u1EE2"

425

" &U << \\u00F9 <<< \\u00D9" /* U */

426

" << \\u1EE7 <<< \\u1EE6"

427

" << \\u0169 <<< \\u0168"

428

" << \\u00FA <<< \\u00DA"

429

" << \\u1EE5 <<< \\u1EE4"

430

" < \\u01B0 <<< \\u01AF" /* U WITH HORN */

431

" << \\u1EEB <<< \\u1EEA"

432

" << \\u1EED <<< \\u1EEC"

433

" << \\u1EEF <<< \\u1EEE"

434

" << \\u1EE9 <<< \\u1EE8"

435

" << \\u1EF1 <<< \\u1EF0"

436

" &Y << \\u1EF3 <<< \\u1EF2" /* Y */

437

" << \\u1EF7 <<< \\u1EF6"

438

" << \\u1EF9 <<< \\u1EF8"

439

" << \\u00FD <<< \\u00DD"

440

" << \\u1EF5 <<< \\u1EF4";

441

442

/* German Phonebook */

443

static const char de_pb_cldr_30[] =

444

"&AE << \\u00E4 <<< \\u00C4 "

445

"&OE << \\u00F6 <<< \\u00D6 "

446

"&UE << \\u00FC <<< \\u00DC ";

447

448

/* Icelandic */

449

static const char is_cldr_30[] =

450

"&[before 1]b < \\u00E1 <<< \\u00C1 "

451

"& d << \\u0111 <<< \\u0110 < \\u00F0 <<< \\u00D0 "

452

"&[before 1]f < \\u00E9 <<< \\u00C9 "

453

"&[before 1]j < \\u00ED <<< \\u00CD "

454

"&[before 1]p < \\u00F3 <<< \\u00D3 "

455

"&[before 1]v < \\u00FA <<< \\u00DA "

456

"&[before 1]z < \\u00FD <<< \\u00DD "

457

"&[before 1]\\u01C0 < \\u00E6 <<< \\u00C6 << \\u00E4 <<< \\u00C4 "

458

"< \\u00F6 <<< \\u00D6 << \\u00F8 <<< \\u00D8 "

459

"< \\u00E5 <<< \\u00C5";

460

461

/* Latvian */

462

static const char lv_cldr_30[] =

463

"&[before 1]D < \\u010D <<< \\u010C "

464

"&[before 1]H < \\u0123 <<< \\u0122 "

465

"& I << y <<< Y "

466

"&[before 1]L < \\u0137 <<< \\u0136 "

467

"&[before 1]M < \\u013C <<< \\u013B "

468

"&[before 1]O < \\u0146 <<< \\u0145 "

469

"&[before 1]S < \\u0157 <<< \\u0156 "

470

"&[before 1]T < \\u0161 <<< \\u0160 "

471

"&[before 1]\\u01B7 < \\u017E <<< \\u017D";

472

473

/* Romanian */

474

static const char ro_cldr_30[] =

475

"&A < \\u0103 <<< \\u0102 < \\u00E2 <<< \\u00C2 "

476

"&I < \\u00EE <<< \\u00CE "

477

"&S < \\u015F = \\u0219 <<< \\u015E = \\u0218 "

478

"&T < \\u0163 = \\u021B <<< \\u0162 = \\u021A";

479

480

/* Slovenian */

481

static const char sl_cldr_30[] =

482

"&C < \\u010D <<< \\u010C < \\u0107 <<< \\u0106 "

483

"&D < \\u0111 <<< \\u0110 "

484

"&S < \\u0161 <<< \\u0160 "

485

"&Z < \\u017E <<< \\u017D";

486

487

/* Polish */

488

static const char pl_cldr_30[] =

489

"&A < \\u0105 <<< \\u0104 "

490

"&C < \\u0107 <<< \\u0106 "

491

"&E < \\u0119 <<< \\u0118 "

492

"&L < \\u0142 <<< \\u0141 "

493

"&N < \\u0144 <<< \\u0143 "

494

"&O < \\u00F3 <<< \\u00D3 "

495

"&S < \\u015B <<< \\u015A "

496

"&Z < \\u017A <<< \\u0179 < \\u017C <<< \\u017B";

497

498

/* Estonian */

499

static const char et_cldr_30[] =

500

"&[before 1]T < \\u0161 <<< \\u0160 < z <<< Z "

501

"< \\u017E <<< \\u017D "

502

"&[before 1]X < \\u00F5 <<< \\u00D5 < \\u00E4 <<< \\u00C4 "

503

"< \\u00F6 <<< \\u00D6 < \\u00FC <<< \\u00DC";

504

505

/* Swedish */

506

static const char sv_cldr_30[] =

507

"& D << \\u0111 <<< \\u0110 << \\u00F0 <<< \\u00D0 "

508

"& t <<< \\u00FE/h "

509

"& T <<< \\u00DE/H "

510

"& Y << \\u00FC <<< \\u00DC << \\u0171 <<< \\u0170 "

511

"&[before 1]\\u01C0 < \\u00E5 <<< \\u00C5 < \\u00E4 <<< \\u00C4 "

512

"<< \\u00E6 <<< \\u00C6 << \\u0119 <<< \\u0118 "

513

"< \\u00F6 <<< \\u00D6 << \\u00F8 <<< \\u00D8 "

514

"<< \\u0151 <<< \\u0150 << \\u0153 <<< \\u0152 "

515

"<< \\u00F4 <<< \\u00D4";

516

517

/* Turkish */

518

static const char tr_cldr_30[] =

519

"& C < \\u00E7 <<< \\u00C7 "

520

"& G < \\u011F <<< \\u011E "

521

"&[before 1]i < \\u0131 <<< I "

522

"& i <<< \\u0130 "

523

"& O < \\u00F6 <<< \\u00D6 "

524

"& S < \\u015F <<< \\u015E "

525

"& U < \\u00FC <<< \\u00DC ";

526

527

/* Czech */

528

static const char cs_cldr_30[] =

529

"&C < \\u010D <<< \\u010C "

530

"&H < ch <<< cH <<< Ch <<< CH "

531

"&R < \\u0159 <<< \\u0158"

532

"&S < \\u0161 <<< \\u0160"

533

"&Z < \\u017E <<< \\u017D";

534

535

/* Danish, same for Norwegian */

536

static const char da_cldr_30[] =

537

"& D << \\u0111 <<< \\u0110 << \\u00F0 <<< \\u00D0 "

538

"& t <<< \\u00FE/h "

539

"& T <<< \\u00DE/H "

540

"& Y << \\u00FC <<< \\u00DC << \\u0171 <<< \\u0170 "

541

"&[before 1]\\u01C0 < \\u00E6 <<< \\u00C6 << \\u00E4 <<< \\u00C4 "

542

"< \\u00F8 <<< \\u00D8 << \\u00F6 <<< \\u00D6 "

543

"<< \\u0151 <<< \\u0150 << \\u0153 <<< \\u0152 "

544

"< \\u00E5 <<< \\u00C5 <<< aa <<< Aa "

545

"<<< AA";

546

547

static Coll_param da_coll_param = {nullptr, false, CASE_FIRST_UPPER};

548

549

/* CASE FIRST OFF for Norwegian */

550

static Coll_param no_coll_param = {nullptr, false, CASE_FIRST_OFF};

551

552

/* Lithuanian */

553

static const char lt_cldr_30[] =

554

"&\\u0300 = \\u0307\\u0300 "

555

"&\\u0301 = \\u0307\\u0301 "

556

"&\\u0303 = \\u0307\\u0303 "

557

"&A << \\u0105 <<< \\u0104 "

558

"&C < \\u010D <<< \\u010C "

559

"&E << \\u0119 <<< \\u0118 << \\u0117 <<< \\u0116"

560

"&I << \\u012F <<< \\u012E << y <<< Y "

561

"&S < \\u0161 <<< \\u0160 "

562

"&U << \\u0173 <<< \\u0172 << \\u016B <<< \\u016A "

563

"&Z < \\u017E <<< \\u017D";

564

565

/* Slovak */

566

static const char sk_cldr_30[] =

567

"&A < \\u00E4 <<< \\u00C4 "

568

"&C < \\u010D <<< \\u010C "

569

"&H < ch <<< cH <<< Ch <<< CH "

570

"&O < \\u00F4 <<< \\u00D4 "

571

"&R < \\u0159 <<< \\u0158 "

572

"&S < \\u0161 <<< \\u0160 "

573

"&Z < \\u017E <<< \\u017D";

574

575

/* Spanish (Traditional) */

576

static const char es_trad_cldr_30[] =

577

"&N < \\u00F1 <<< \\u00D1 "

578

"&C < ch <<< Ch <<< CH "

579

"&l < ll <<< Ll <<< LL";

/* Persian */

#if 0

static const char fa_cldr_30[]=

584

"& \\u064E << \\u0650 << \\u064F << \\u064B << \\u064D "

585

"<< \\u064C "

586

"&[before 1]\\u0627 < \\u0622 "

587

"& \\u0627 << \\u0671 < \\u0621 << \\u0623 << \\u0672 "

588

"<< \\u0625 << \\u0673 << \\u0624 << \\u06CC\\u0654 "

589

"<<< \\u0649\\u0654 <<< \\u0626 "

590

"& \\u06A9 << \\u06AA << \\u06AB << \\u0643 << \\u06AC "

591

"<< \\u06AD << \\u06AE "

592

"& \\u06CF < \\u0647 << \\u06D5 << \\u06C1 << \\u0629 "

593

"<< \\u06C3 << \\u06C0 << \\u06BE "

594

"& \\u06CC << \\u0649 << \\u06D2 << \\u064A << \\u06D0 "

595

"<< \\u06D1 << \\u06CD << \\u06CE";

596

597

static Reorder_param fa_reorder_param= {

598

{CHARGRP_ARAB, CHARGRP_NONE}, {{{0, 0}, {0, 0}}}, 0

599

};

600

601

static Coll_param fa_coll_param= {

602

&fa_reorder_param, true

};

#endif

/* Hungarian */

static const char hu_cldr_30[] =

608

"&C < cs <<< Cs <<< CS "

609

"&D < dz <<< Dz <<< DZ "

610

"&DZ < dzs <<< Dzs <<< DZS "

611

"&G < gy <<< Gy <<< GY "

612

"&L < ly <<< Ly <<< LY "

613

"&N < ny <<< Ny <<< NY "

614

"&S < sz <<< Sz <<< SZ "

615

"&T < ty <<< Ty <<< TY "

616

"&Z < zs <<< Zs <<< ZS "

617

"&O < \\u00F6 <<< \\u00D6 << \\u0151 <<< \\u0150 "

618

"&U < \\u00FC <<< \\u00DC << \\u0171 <<< \\u0170 "

"&cs <<< ccs/cs "

"&Cs <<< Ccs/cs "

"&CS <<< CCS/CS "

"&dz <<< ddz/dz "

"&Dz <<< Ddz/dz "

"&DZ <<< DDZ/DZ "

"&dzs<<< ddzs/dzs "

"&Dzs<<< Ddzs/dzs "

"&DZS<<< DDZS/DZS "

"&gy <<< ggy/gy "

"&Gy <<< Ggy/gy "

"&GY <<< GGY/GY "

"&ly <<< lly/ly "

"&Ly <<< Lly/ly "

"&LY <<< LLY/LY "

"&ny <<< nny/ny "

"&Ny <<< Nny/ny "

"&NY <<< NNY/NY "

"&sz <<< ssz/sz "

"&Sz <<< Ssz/sz "

"&SZ <<< SSZ/SZ "

"&ty <<< tty/ty "

"&Ty <<< Tty/ty "

"&TY <<< TTY/TY "

"&zs <<< zzs/zs "

"&Zs <<< Zzs/zs "

"&ZS <<< ZZS/ZS";

/* Croatian, same for Serbian with Latin and Bosnian. */

648

static const char hr_cldr_30[] =

649

"&C < \\u010D <<< \\u010C < \\u0107 <<< \\u0106 "

650

"&D < d\\u017E <<< \\u01C6 <<< D\\u017E <<< \\u01C5 <<< D\\u017D "

651

"<<< \\u01C4 < \\u0111 <<< \\u0110 "

652

"&L < lj <<< \\u01C9 <<< Lj <<< \\u01C8 <<< LJ "

653

"<<< \\u01C7 "

654

"&N < nj <<< \\u01CC <<< Nj <<< \\u01CB <<< NJ "

655

"<<< \\u01CA "

656

"&S < \\u0161 <<< \\u0160 "

657

"&Z < \\u017E <<< \\u017D ";

658

659

static Reorder_param hr_reorder_param = {

660

{CHARGRP_LATIN, CHARGRP_CYRILLIC, CHARGRP_NONE}, {{{0, 0}, {0, 0}}}, 0, 0};

661

662

static Coll_param hr_coll_param = {&hr_reorder_param, false, CASE_FIRST_OFF};

/* Sinhala */

#if 0

static const char si_cldr_30[]=

667

"&\\u0D96 < \\u0D82 < \\u0D83 "

668

"&\\u0DA5 < \\u0DA4";

#endif

/* Vietnamese */

static const char vi_cldr_30[] =

673

"&\\u0300 << \\u0309 << \\u0303 << \\u0301 << \\u0323 "

674

"&a < \\u0103 <<< \\u0102 < \\u00E2 <<< \\u00C2 "

675

"&d < \\u0111 <<< \\u0110 "

676

"&e < \\u00EA <<< \\u00CA "

677

"&o < \\u00F4 <<< \\u00D4 < \\u01A1 <<< \\u01A0 "

678

"&u < \\u01B0 <<< \\u01AF";

679

680

static Coll_param vi_coll_param = {nullptr, true, CASE_FIRST_OFF};

681

682

static Reorder_param ja_reorder_param = {

683

/*

684

Per CLDR 30, Japanese reorder rule is defined as [Latn Kana Hani],

685

but for Hani characters, their weight is implicit according to UCA,

686

which is different from other character groups. We don't add "Hani"

687

below and will have special handling for them in

688

adjust_japanese_weight() and apply_reorder_param(). Implicit weight

689

has two collation elements. To make strnxfrm() run faster, we give

690

Japanese Han characters tailored weight which has only one collation

691

element. These characters' weight is defined in ja_han_pages.

692

*/

693

{CHARGRP_LATIN, CHARGRP_KANA, CHARGRP_NONE},

{{{0, 0}, {0, 0}}},

0,

0};

static Coll_param ja_coll_param = {&ja_reorder_param, false /*norm_enabled*/,

CASE_FIRST_OFF};

/*

The Chinese reorder rule is defined as [Hani]. This means all Han characters'

703

weight should be greater than the core group and smaller than any other

704

character groups.

705

The Han characters are separated into two parts. The CLDR collation

706

definition file, zh.xml, defines 41336 Han characters' order, and all other

707

Han characters have implicit weight.

708

Since the core group characters occupy the weight value from 0x0209 to 0x1C46

709

in DUCET, so we decide to set the weight of all Han characters defined in

710

zh.xml to be the value from 0x1C47 to 0xBDBE. The smallest weight value of

711

these Han characters, 0x1C47, being the largest weight value of the core

712

group plus one (0x1C46 + 1), ensures these Han characters sort greater than

713

the core group characters.

714

Also, we set the implicit weight to the Han characters like

715

[BDBF - BDC3, 0020, 0002][XXXX, 0000, 0000].

716

To tailor the weight of characters of Latin, Cyrillic and so on to be bigger

717

than all Han characters, we give these characters weights from 0xBDC4 to

718

0xF620. There are many character groups between the core group and the Han

719

group, so it would be a long list if we put them in the following reorder_grp

720

structure. But since it is a very simple weight shift, we put their calculated

721

weight here and do not calculate it in my_prepare_reorder().

722

723

NOTE: We use the zh.xml file from CLDR v33.1 to implement this Chinese

724

collation, because we found that the file of CLDR v30 is missing some very

725

common Han characters (the Han character 'small', etc).

726

*/

727

static Reorder_param zh_reorder_param = {

728

{CHARGRP_NONE}, {{{0x1C47, 0x54A3}, {0xBDC4, 0xF620}}}, 1, 0x54A3};

729

730

static Coll_param zh_coll_param = {&zh_reorder_param, false, CASE_FIRST_OFF};

731

732

/* Russian, same for Bulgerian and Mongolian with Cyrillic letters */

733

static Reorder_param ru_reorder_param = {

734

{CHARGRP_CYRILLIC, CHARGRP_NONE}, {{{0, 0}, {0, 0}}}, 0, 0};

735

736

static Coll_param ru_coll_param = {&ru_reorder_param, false /*norm_enabled*/,

737

CASE_FIRST_OFF};

738

739

static constexpr uint16 nochar[] = {0, 0};

740

741

/**

742

Unicode Collation Algorithm:

743

Collation element (weight) scanner,

744

for consequent scan of collations

745

weights from a string.

746

747

Only meant as a base class; instantiate uca_scanner_any or uca_scanner_900

748

instead of this.

749

*/

750

class my_uca_scanner {

751

protected:

752

11076031406

my_uca_scanner(const CHARSET_INFO *cs_arg, const uchar *str, size_t length)

753

11076031406

: wbeg(nochar),

754

11076031406

sbeg(str),

755

11076031406

send(str + length),

756

11076031406

uca(cs_arg->uca),

757

11076031406

cs(cs_arg),

758

11076031406

sbeg_dup(str) {}

public:

/**

Get the level the scanner is currently working on. The string

763

can be scanned multiple times (if the collation requires multi-level

764

comparisons, e.g. for accent or case sensitivity); first to get

765

primary weights, then from the start again for secondary, etc.

766

*/

767

38020806387

uint get_weight_level() const { return weight_lv; }

768

769

protected:

770

uint weight_lv{0}; /* 0 = Primary, 1 = Secondary, 2 = Tertiary */

771

const uint16 *wbeg; /* Beginning of the current weight string */

772

uint wbeg_stride{0}; /* Number of bytes between weights in string */

773

const uchar *sbeg; /* Beginning of the input string */

774

const uchar *send; /* End of the input string */

775

const MY_UCA_INFO *uca;

776

uint16 implicit[10];

777

my_wc_t prev_char{0}; // Previous code point we scanned, if any.

778

const CHARSET_INFO *cs;

779

uint num_of_ce_left{0};

780

const uchar *sbeg_dup; /* Backup of beginning of input string */

781

782

protected:

783

const uint16 *contraction_find(my_wc_t wc0, size_t *chars_skipped);

784

inline const uint16 *previous_context_find(my_wc_t wc0, my_wc_t wc1);

};

/*

Charset dependent scanner part, to optimize

some character sets.

*/

template <class Mb_wc>

793

struct uca_scanner_any : public my_uca_scanner {

794

508351708

uca_scanner_any(const Mb_wc mb_wc, const CHARSET_INFO *cs_arg,

795

const uchar *str, size_t length)

796

508351708

: my_uca_scanner(cs_arg, str, length), mb_wc(mb_wc) {

797

// UCA 9.0.0 uses a different table format from what this scanner expects.

798

2/4

✓ Branch 0 taken 254175871 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 254175871 times.

508351724

assert(cs_arg->uca == nullptr || cs_arg->uca->version != UCA_V900);

799

508351724

}

800

801

1014477204

uint get_char_index() const { return char_index; }

inline int next();

private:

/**

How many code points (possibly multibyte) we have scanned so far.

808

This includes code points with zero weight. Note that this is reset

809

once we get to the end of the string and restart the scanning for

810

the next weight level, but it is _not_ reset when we reach the

811

end of the last level.

*/

uint char_index{0};

const Mb_wc mb_wc;

inline int next_implicit(my_wc_t ch);

818

};

819

820

template <class Mb_wc, int LEVELS_FOR_COMPARE>

821

class uca_scanner_900 : public my_uca_scanner {

822

public:

823

21643811440

uca_scanner_900(const Mb_wc mb_wc, const CHARSET_INFO *cs_arg,

824

const uchar *str, size_t length)

825

21643811440

: my_uca_scanner(cs_arg, str, length), mb_wc(mb_wc) {}

inline int next();

/**

For each weight in sequence, call "func", which should have

831

a function signature of "bool func(int weight, bool is_level_separator)".

832

Stops the iteration early if "func" returns false.

833

834

This is morally equivalent to

835

836

int weight;

837

while ((weight= next()) >= 0)

838

{

839

if (!func(weight, weight == 0)) break;

840

}

841

842

except that it might employ optimizations internally to speed up

843

the process. These optimizations will not modify the number of calls

844

to func() (or their order), but might affect the internal scanner

845

state during the calls, so func() should not try to read from

846

the scanner except by calling public member functions.

847

848

As a special optimization, if "bool preaccept_data(int num_weights)"

849

returns true, the next "num_weights" calls to func() _must_ return

850

true. This is so that bounds checking costs can be amortized

851

over fewer calls.

852

*/

853

template <class T, class U>

854

inline void for_each_weight(T func, U preaccept_data);

private:

const Mb_wc mb_wc;

inline int next_raw();

860

inline int more_weight();

861

uint16 apply_case_first(uint16 weight);

862

uint16 apply_reorder_param(uint16 weight);

863

inline int next_implicit(my_wc_t ch);

864

void my_put_jamo_weights(my_wc_t *hangul_jamo, int jamo_cnt);

865

/*

866

apply_reorder_param() needs to return two weights for each origin

867

weight. This boolean signals whether we have already returned the

868

FB86 weight, and are ready to return the origin weight.

869

*/

870

bool return_origin_weight{true};

871

/*

872

For Japanese kana-sensitive collation, we only add quaternary

873

weight for katakana and hiragana, but not for others like latin

874

and kanji, because characters like latin and kanji can be already

875

distinguished from kana by three levels of weight.

876

has_quaternary_weight is to indicate whether quaternary weight is

877

needed for characters in string.

878

*/

879

bool has_quaternary_weight{false};

880

int handle_ja_contraction_quat_wt();

881

int handle_ja_common_quat_wt(my_wc_t wc);

882

};

883

884

/********** Helper functions to handle contraction ************/

885

886

/**

887

Mark a code point as a contraction part

888

889

@param flags Pointer to UCA contraction flag data

890

@param wc Unicode code point

891

@param flag flag: "is contraction head", "is contraction tail"

892

*/

893

894

546791

static inline void my_uca_add_contraction_flag(char *flags, my_wc_t wc,

895

int flag) {

896

546791

flags[wc & MY_UCA_CNT_FLAG_MASK] |= flag;

897

546791

}

898

899

/**

900

Check if UCA level data has contractions.

901

902

@param uca Pointer to UCA data

903

904

@return Flags indicating if UCA with contractions

905

@retval 0 - no contractions

906

@retval 1 - there are some contractions

907

*/

908

909

37803780609

static inline bool my_uca_have_contractions(const MY_UCA_INFO *uca) {

910

37803780609

return uca->have_contractions;

911

}

912

913

struct trie_node_cmp {

914

3021597

bool operator()(const MY_CONTRACTION &a, const my_wc_t b) { return a.ch < b; }

915

bool operator()(const MY_CONTRACTION &a, const MY_CONTRACTION &b) {

return a.ch < b.ch;

}

};

static std::vector<MY_CONTRACTION>::const_iterator

921

489459

find_contraction_part_in_trie(const std::vector<MY_CONTRACTION> &cont_nodes,

922

my_wc_t ch) {

923

2/2

✓ Branch 0 taken 21062 times.

✓ Branch 1 taken 468397 times.

489459

if (cont_nodes.empty()) return cont_nodes.end();

924

468397

return std::lower_bound(cont_nodes.begin(), cont_nodes.end(), ch,

925

468397

trie_node_cmp());

926

}

927

928

549253

static std::vector<MY_CONTRACTION>::iterator find_contraction_part_in_trie(

929

std::vector<MY_CONTRACTION> &cont_nodes, my_wc_t ch) {

930

2/2

✓ Branch 0 taken 52185 times.

✓ Branch 1 taken 497068 times.

549253

if (cont_nodes.empty()) return cont_nodes.end();

931

497068

return std::lower_bound(cont_nodes.begin(), cont_nodes.end(), ch,

932

497068

trie_node_cmp());

933

}

934

/**

935

Find a contraction consisting of two code points and return its weight array

936

937

@param cont_nodes Vector that contains contraction nodes

938

@param wc1 First code point

939

@param wc2 Second code point

940

941

@return Weight array

942

@retval NULL - no contraction found

943

@retval ptr - contraction weight array

944

*/

945

946

199

const uint16 *my_uca_contraction2_weight(

947

const std::vector<MY_CONTRACTION> *cont_nodes, my_wc_t wc1, my_wc_t wc2) {

948

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 199 times.

199

if (!cont_nodes) return nullptr;

949

950

1/2

✓ Branch 0 taken 199 times.

✗ Branch 1 not taken.

199

if (!cont_nodes->empty()) {

951

std::vector<MY_CONTRACTION>::const_iterator node_it1 =

952

1/2

✓ Branch 0 taken 199 times.

✗ Branch 1 not taken.

199

find_contraction_part_in_trie(*cont_nodes, wc1);

953

3/6

✓ Branch 0 taken 199 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 199 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 199 times.

398

if (node_it1 == cont_nodes->end() || node_it1->ch != wc1) return nullptr;

954

std::vector<MY_CONTRACTION>::const_iterator node_it2 =

955

1/2

✓ Branch 0 taken 199 times.

✗ Branch 1 not taken.

199

find_contraction_part_in_trie(node_it1->child_nodes, wc2);

956

3/6

✓ Branch 0 taken 199 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 199 times.

✗ Branch 3 not taken.

✓ Branch 4 taken 199 times.

✗ Branch 5 not taken.

398

if (node_it2 != node_it1->child_nodes.end() && node_it2->ch == wc2 &&

957

1/2

✓ Branch 0 taken 199 times.

✗ Branch 1 not taken.

199

node_it2->is_contraction_tail)

958

199

return node_it2->weight;

959

}

960

✗

return nullptr;

}

/**

Check if a code point can be previous context head

965

966

@param flags Pointer to UCA contraction flag data

967

@param wc Code point

968

969

@retval false - cannot be previous context head

970

@retval true - can be previous context head

971

*/

972

973

27040

static inline bool my_uca_can_be_previous_context_head(const char *flags,

974

my_wc_t wc) {

975

27040

return flags[wc & MY_UCA_CNT_FLAG_MASK] & MY_UCA_PREVIOUS_CONTEXT_HEAD;

}

/**

Check if a code point can be previous context tail

980

981

@param flags Pointer to UCA contraction flag data

982

@param wc Code point

983

984

@retval false - cannot be contraction tail

985

@retval true - can be contraction tail

986

*/

987

988

182990785

static inline bool my_uca_can_be_previous_context_tail(const char *flags,

989

my_wc_t wc) {

990

182990785

return flags[wc & MY_UCA_CNT_FLAG_MASK] & MY_UCA_PREVIOUS_CONTEXT_TAIL;

}

/**

Check if a string is a contraction of exactly the given length,

995

and return its weight array on success.

996

997

@param cont_nodes Vector that contains contraction nodes

998

@param wc Pointer to wide string

999

@param len String length

1000

1001

@return Weight array

1002

@retval NULL - Input string is not a known contraction

1003

@retval ptr - contraction weight array

1004

*/

1005

1006

376988

static inline const uint16 *my_uca_contraction_weight(

1007

const std::vector<MY_CONTRACTION> *cont_nodes, const my_wc_t *wc,

1008

size_t len) {

1009

2/2

✓ Branch 0 taken 180420 times.

✓ Branch 1 taken 196568 times.

376988

if (!cont_nodes) return nullptr;

1010

1011

196568

std::vector<MY_CONTRACTION>::const_iterator node_it;

1012

2/2

✓ Branch 0 taken 270557 times.

✓ Branch 1 taken 20987 times.

291544

for (size_t ch_ind = 0; ch_ind < len; ++ch_ind) {

1013

1/2

✓ Branch 0 taken 270557 times.

✗ Branch 1 not taken.

270557

node_it = find_contraction_part_in_trie(*cont_nodes, wc[ch_ind]);

1014

6/6

✓ Branch 0 taken 200743 times.

✓ Branch 1 taken 69814 times.

✓ Branch 2 taken 105767 times.

✓ Branch 3 taken 94976 times.

✓ Branch 4 taken 175581 times.

✓ Branch 5 taken 94976 times.

270557

if (node_it == cont_nodes->end() || node_it->ch != wc[ch_ind])

1015

175581

return nullptr;

1016

94976

cont_nodes = &node_it->child_nodes;

1017

}

1018

2/2

✓ Branch 0 taken 20027 times.

✓ Branch 1 taken 960 times.

20987

if (node_it->is_contraction_tail) return node_it->weight;

1019

960

return nullptr;

}

/**

Return length of a 0-terminated wide string, analogous to strnlen().

1024

1025

@param s Pointer to wide string

1026

@param maxlen Mamixum string length

1027

1028

@return string length, or maxlen if no '\0' is met.

1029

*/

1030

2813640

static size_t my_wstrnlen(my_wc_t *s, size_t maxlen) {

1031

2/2

✓ Branch 0 taken 6517219 times.

✓ Branch 1 taken 1081 times.

6518300

for (size_t i = 0; i < maxlen; i++) {

1032

2/2

✓ Branch 0 taken 2812559 times.

✓ Branch 1 taken 3704660 times.

6517219

if (s[i] == 0) return i;

1033

}

1034

1081

return maxlen;

}

/**

Find a contraction in the input stream and return its weight array

1039

1040

Scan input code points to find a longest path in contraction trie

1041

which contains all these code points. If the ending node of this

1042

path is end of contraction, return the weight array.

1043

1044

@param wc0 The first code point of the contraction (which should have

1045

the MY_UCA_CNT_HEAD flag).

1046

@param[out] chars_skipped How many code points where skipped in the

1047

contraction we found. Only makes sense if we actually found one.

1048

1049

@return Weight array

1050

@retval NULL no contraction found

1051

@retval ptr contraction weight array

1052

*/

1053

1054

201097

const uint16 *my_uca_scanner::contraction_find(my_wc_t wc0,

1055

size_t *chars_skipped) {

1056

201097

const uchar *beg = nullptr;

1057

201097

auto mb_wc = cs->cset->mb_wc;

1058

1059

201097

const uchar *s = sbeg;

1060

201097

const std::vector<MY_CONTRACTION> *cont_nodes = uca->contraction_nodes;

1061

201097

const MY_CONTRACTION *longest_contraction = nullptr;

1062

201097

std::vector<MY_CONTRACTION>::const_iterator node_it;

1063

for (;;) {

1064

1/2

✓ Branch 0 taken 216042 times.

✗ Branch 1 not taken.

216042

node_it = find_contraction_part_in_trie(*cont_nodes, wc0);

1065

6/6

✓ Branch 0 taken 26844 times.

✓ Branch 1 taken 189198 times.

✓ Branch 2 taken 6476 times.

✓ Branch 3 taken 20368 times.

✓ Branch 4 taken 195674 times.

✓ Branch 5 taken 20368 times.

216042

if (node_it == cont_nodes->end() || node_it->ch != wc0) break;

1066

2/2

✓ Branch 0 taken 3734 times.

✓ Branch 1 taken 16634 times.

20368

if (node_it->is_contraction_tail) {

1067

3734

longest_contraction = &(*node_it);

1068

3734

beg = s;

1069

3734

*chars_skipped = node_it->contraction_len - 1;

1070

}

1071

int mblen;

1072

3/4

✓ Branch 0 taken 20368 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 5423 times.

✓ Branch 3 taken 14945 times.

20368

if ((mblen = mb_wc(cs, &wc0, s, send)) <= 0) break;

1073

14945

s += mblen;

1074

14945

cont_nodes = &node_it->child_nodes;

1075

14945

}

1076

1077

2/2

✓ Branch 0 taken 3509 times.

✓ Branch 1 taken 197588 times.

201097

if (longest_contraction != nullptr) {

1078

3509

const uint16 *cweight = longest_contraction->weight;

1079

2/2

✓ Branch 0 taken 890 times.

✓ Branch 1 taken 2619 times.

3509

if (uca->version == UCA_V900) {

1080

890

cweight += weight_lv;

1081

890

wbeg = cweight + MY_UCA_900_CE_SIZE;

1082

890

wbeg_stride = MY_UCA_900_CE_SIZE;

1083

890

num_of_ce_left = 7;

1084

} else {

1085

2619

wbeg = cweight + 1;

1086

2619

wbeg_stride = MY_UCA_900_CE_SIZE;

1087

}

1088

3509

sbeg = beg;

1089

3509

return cweight;

1090

}

1091

197588

return nullptr; /* No contractions were found */

}

/**

Find weight for contraction with previous context

1096

and return its weight array.

1097

1098

@param wc0 Previous code point

1099

@param wc1 Current code point

1100

1101

@return Weight array

1102

@retval NULL - no contraction with context found

1103

@retval ptr - contraction weight array

1104

*/

1105

ALWAYS_INLINE

1106

const uint16 *my_uca_scanner::previous_context_find(my_wc_t wc0, my_wc_t wc1) {

1107

std::vector<MY_CONTRACTION>::const_iterator node_it1 =

1108

2462

find_contraction_part_in_trie(*uca->contraction_nodes, wc1);

1109

18/444

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 1613 times.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✓ Branch 51 taken 1613 times.

✗ Branch 52 not taken.

✓ Branch 53 taken 1613 times.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✓ Branch 60 taken 639 times.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✓ Branch 63 taken 639 times.

✗ Branch 64 not taken.

✓ Branch 65 taken 639 times.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✓ Branch 144 taken 56 times.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✓ Branch 147 taken 56 times.

✗ Branch 148 not taken.

✓ Branch 149 taken 56 times.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✗ Branch 213 not taken.

✗ Branch 214 not taken.

✗ Branch 215 not taken.

✗ Branch 216 not taken.

✗ Branch 217 not taken.

✗ Branch 218 not taken.

✗ Branch 219 not taken.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

✗ Branch 224 not taken.

✗ Branch 225 not taken.

✗ Branch 226 not taken.

✗ Branch 227 not taken.

✗ Branch 228 not taken.

✗ Branch 229 not taken.

✗ Branch 230 not taken.

✗ Branch 231 not taken.

✗ Branch 232 not taken.

✗ Branch 233 not taken.

✗ Branch 234 not taken.

✗ Branch 235 not taken.

✗ Branch 236 not taken.

✗ Branch 237 not taken.

✗ Branch 238 not taken.

✗ Branch 239 not taken.

✗ Branch 240 not taken.

✗ Branch 241 not taken.

✗ Branch 242 not taken.

✗ Branch 243 not taken.

✗ Branch 244 not taken.

✗ Branch 245 not taken.

✗ Branch 246 not taken.

✗ Branch 247 not taken.

✗ Branch 248 not taken.

✗ Branch 249 not taken.

✗ Branch 250 not taken.

✗ Branch 251 not taken.

✗ Branch 252 not taken.

✗ Branch 253 not taken.

✗ Branch 254 not taken.

✗ Branch 255 not taken.

✗ Branch 256 not taken.

✗ Branch 257 not taken.

✗ Branch 258 not taken.

✗ Branch 259 not taken.

✗ Branch 260 not taken.

✗ Branch 261 not taken.

✗ Branch 262 not taken.

✗ Branch 263 not taken.

✓ Branch 264 taken 57 times.

✗ Branch 265 not taken.

✗ Branch 266 not taken.

✓ Branch 267 taken 57 times.

✗ Branch 268 not taken.

✓ Branch 269 taken 57 times.

✓ Branch 270 taken 57 times.

✗ Branch 271 not taken.

✗ Branch 272 not taken.

✓ Branch 273 taken 57 times.

✗ Branch 274 not taken.

✓ Branch 275 taken 57 times.

✗ Branch 276 not taken.

✗ Branch 277 not taken.

✗ Branch 278 not taken.

✗ Branch 279 not taken.

✗ Branch 280 not taken.

✗ Branch 281 not taken.

✗ Branch 282 not taken.

✗ Branch 283 not taken.

✗ Branch 284 not taken.

✗ Branch 285 not taken.

✗ Branch 286 not taken.

✗ Branch 287 not taken.

✗ Branch 288 not taken.

✗ Branch 289 not taken.

✗ Branch 290 not taken.

✗ Branch 291 not taken.

✗ Branch 292 not taken.

✗ Branch 293 not taken.

✗ Branch 294 not taken.

✗ Branch 295 not taken.

✗ Branch 296 not taken.

✗ Branch 297 not taken.

✗ Branch 298 not taken.

✗ Branch 299 not taken.

✗ Branch 300 not taken.

✗ Branch 301 not taken.

✗ Branch 302 not taken.

✗ Branch 303 not taken.

✗ Branch 304 not taken.

✗ Branch 305 not taken.

✗ Branch 306 not taken.

✗ Branch 307 not taken.

✗ Branch 308 not taken.

✗ Branch 309 not taken.

✗ Branch 310 not taken.

✗ Branch 311 not taken.

✗ Branch 312 not taken.

✗ Branch 313 not taken.

✗ Branch 314 not taken.

✗ Branch 315 not taken.

✗ Branch 316 not taken.

✗ Branch 317 not taken.

✗ Branch 318 not taken.

✗ Branch 319 not taken.

✗ Branch 320 not taken.

✗ Branch 321 not taken.

✗ Branch 322 not taken.

✗ Branch 323 not taken.

✗ Branch 324 not taken.

✗ Branch 325 not taken.

✗ Branch 326 not taken.

✗ Branch 327 not taken.

✗ Branch 328 not taken.

✗ Branch 329 not taken.

✗ Branch 330 not taken.

✗ Branch 331 not taken.

✗ Branch 332 not taken.

✗ Branch 333 not taken.

✗ Branch 334 not taken.

✗ Branch 335 not taken.

✓ Branch 336 taken 40 times.

✗ Branch 337 not taken.

✗ Branch 338 not taken.

✓ Branch 339 taken 40 times.

✗ Branch 340 not taken.

✓ Branch 341 taken 40 times.

✗ Branch 342 not taken.

✗ Branch 343 not taken.

✗ Branch 344 not taken.

✗ Branch 345 not taken.

✗ Branch 346 not taken.

✗ Branch 347 not taken.

✗ Branch 348 not taken.

✗ Branch 349 not taken.

✗ Branch 350 not taken.

✗ Branch 351 not taken.

✗ Branch 352 not taken.

✗ Branch 353 not taken.

✗ Branch 354 not taken.

✗ Branch 355 not taken.

✗ Branch 356 not taken.

✗ Branch 357 not taken.

✗ Branch 358 not taken.

✗ Branch 359 not taken.

✗ Branch 360 not taken.

✗ Branch 361 not taken.

✗ Branch 362 not taken.

✗ Branch 363 not taken.

✗ Branch 364 not taken.

✗ Branch 365 not taken.

✗ Branch 366 not taken.

✗ Branch 367 not taken.

✗ Branch 368 not taken.

✗ Branch 369 not taken.

✗ Branch 370 not taken.

✗ Branch 371 not taken.

✗ Branch 372 not taken.

✗ Branch 373 not taken.

✗ Branch 374 not taken.

✗ Branch 375 not taken.

✗ Branch 376 not taken.

✗ Branch 377 not taken.

✗ Branch 378 not taken.

✗ Branch 379 not taken.

✗ Branch 380 not taken.

✗ Branch 381 not taken.

✗ Branch 382 not taken.

✗ Branch 383 not taken.

✗ Branch 384 not taken.

✗ Branch 385 not taken.

✗ Branch 386 not taken.

✗ Branch 387 not taken.

✗ Branch 388 not taken.

✗ Branch 389 not taken.

✗ Branch 390 not taken.

✗ Branch 391 not taken.

✗ Branch 392 not taken.

✗ Branch 393 not taken.

✗ Branch 394 not taken.

✗ Branch 395 not taken.

✗ Branch 396 not taken.

✗ Branch 397 not taken.

✗ Branch 398 not taken.

✗ Branch 399 not taken.

✗ Branch 400 not taken.

✗ Branch 401 not taken.

✗ Branch 402 not taken.

✗ Branch 403 not taken.

✗ Branch 404 not taken.

✗ Branch 405 not taken.

✗ Branch 406 not taken.

✗ Branch 407 not taken.

✗ Branch 408 not taken.

✗ Branch 409 not taken.

✗ Branch 410 not taken.

✗ Branch 411 not taken.

✗ Branch 412 not taken.

✗ Branch 413 not taken.

✗ Branch 414 not taken.

✗ Branch 415 not taken.

✗ Branch 416 not taken.

✗ Branch 417 not taken.

✗ Branch 418 not taken.

✗ Branch 419 not taken.

✗ Branch 420 not taken.

✗ Branch 421 not taken.

✗ Branch 422 not taken.

✗ Branch 423 not taken.

✗ Branch 424 not taken.

✗ Branch 425 not taken.

✗ Branch 426 not taken.

✗ Branch 427 not taken.

✗ Branch 428 not taken.

✗ Branch 429 not taken.

✗ Branch 430 not taken.

✗ Branch 431 not taken.

✗ Branch 432 not taken.

✗ Branch 433 not taken.

✗ Branch 434 not taken.

✗ Branch 435 not taken.

✗ Branch 436 not taken.

✗ Branch 437 not taken.

✗ Branch 438 not taken.

✗ Branch 439 not taken.

✗ Branch 440 not taken.

✗ Branch 441 not taken.

✗ Branch 442 not taken.

✗ Branch 443 not taken.

2462

if (node_it1 == uca->contraction_nodes->end() || node_it1->ch != wc1)

1110

✗

return nullptr;

1111

std::vector<MY_CONTRACTION>::const_iterator node_it2 =

1112

6/148

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 1613 times.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 639 times.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 56 times.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 57 times.

✗ Branch 89 not taken.

✓ Branch 90 taken 57 times.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✓ Branch 112 taken 40 times.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

2462

find_contraction_part_in_trie(node_it1->child_nodes_context, wc0);

1113

18/444

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 1613 times.

✗ Branch 49 not taken.

✓ Branch 50 taken 1613 times.

✗ Branch 51 not taken.

✓ Branch 52 taken 1613 times.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✓ Branch 60 taken 639 times.

✗ Branch 61 not taken.

✓ Branch 62 taken 639 times.

✗ Branch 63 not taken.

✓ Branch 64 taken 639 times.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✓ Branch 144 taken 56 times.

✗ Branch 145 not taken.

✓ Branch 146 taken 56 times.

✗ Branch 147 not taken.

✓ Branch 148 taken 56 times.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✗ Branch 213 not taken.

✗ Branch 214 not taken.

✗ Branch 215 not taken.

✗ Branch 216 not taken.

✗ Branch 217 not taken.

✗ Branch 218 not taken.

✗ Branch 219 not taken.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

✗ Branch 224 not taken.

✗ Branch 225 not taken.

✗ Branch 226 not taken.

✗ Branch 227 not taken.

✗ Branch 228 not taken.

✗ Branch 229 not taken.

✗ Branch 230 not taken.

✗ Branch 231 not taken.

✗ Branch 232 not taken.

✗ Branch 233 not taken.

✗ Branch 234 not taken.

✗ Branch 235 not taken.

✗ Branch 236 not taken.

✗ Branch 237 not taken.

✗ Branch 238 not taken.

✗ Branch 239 not taken.

✗ Branch 240 not taken.

✗ Branch 241 not taken.

✗ Branch 242 not taken.

✗ Branch 243 not taken.

✗ Branch 244 not taken.

✗ Branch 245 not taken.

✗ Branch 246 not taken.

✗ Branch 247 not taken.

✗ Branch 248 not taken.

✗ Branch 249 not taken.

✗ Branch 250 not taken.

✗ Branch 251 not taken.

✗ Branch 252 not taken.

✗ Branch 253 not taken.

✗ Branch 254 not taken.

✗ Branch 255 not taken.

✗ Branch 256 not taken.

✗ Branch 257 not taken.

✗ Branch 258 not taken.

✗ Branch 259 not taken.

✗ Branch 260 not taken.

✗ Branch 261 not taken.

✗ Branch 262 not taken.

✗ Branch 263 not taken.

✓ Branch 264 taken 57 times.

✗ Branch 265 not taken.

✓ Branch 266 taken 57 times.

✗ Branch 267 not taken.

✓ Branch 268 taken 57 times.

✗ Branch 269 not taken.

✓ Branch 270 taken 57 times.

✗ Branch 271 not taken.

✓ Branch 272 taken 57 times.

✗ Branch 273 not taken.

✓ Branch 274 taken 57 times.

✗ Branch 275 not taken.

✗ Branch 276 not taken.

✗ Branch 277 not taken.

✗ Branch 278 not taken.

✗ Branch 279 not taken.

✗ Branch 280 not taken.

✗ Branch 281 not taken.

✗ Branch 282 not taken.

✗ Branch 283 not taken.

✗ Branch 284 not taken.

✗ Branch 285 not taken.

✗ Branch 286 not taken.

✗ Branch 287 not taken.

✗ Branch 288 not taken.

✗ Branch 289 not taken.

✗ Branch 290 not taken.

✗ Branch 291 not taken.

✗ Branch 292 not taken.

✗ Branch 293 not taken.

✗ Branch 294 not taken.

✗ Branch 295 not taken.

✗ Branch 296 not taken.

✗ Branch 297 not taken.

✗ Branch 298 not taken.

✗ Branch 299 not taken.

✗ Branch 300 not taken.

✗ Branch 301 not taken.

✗ Branch 302 not taken.

✗ Branch 303 not taken.

✗ Branch 304 not taken.

✗ Branch 305 not taken.

✗ Branch 306 not taken.

✗ Branch 307 not taken.

✗ Branch 308 not taken.

✗ Branch 309 not taken.

✗ Branch 310 not taken.

✗ Branch 311 not taken.

✗ Branch 312 not taken.

✗ Branch 313 not taken.

✗ Branch 314 not taken.

✗ Branch 315 not taken.

✗ Branch 316 not taken.

✗ Branch 317 not taken.

✗ Branch 318 not taken.

✗ Branch 319 not taken.

✗ Branch 320 not taken.

✗ Branch 321 not taken.

✗ Branch 322 not taken.

✗ Branch 323 not taken.

✗ Branch 324 not taken.

✗ Branch 325 not taken.

✗ Branch 326 not taken.

✗ Branch 327 not taken.

✗ Branch 328 not taken.

✗ Branch 329 not taken.

✗ Branch 330 not taken.

✗ Branch 331 not taken.

✗ Branch 332 not taken.

✗ Branch 333 not taken.

✗ Branch 334 not taken.

✗ Branch 335 not taken.

✓ Branch 336 taken 40 times.

✗ Branch 337 not taken.

✓ Branch 338 taken 40 times.

✗ Branch 339 not taken.

✓ Branch 340 taken 40 times.

✗ Branch 341 not taken.

✗ Branch 342 not taken.

✗ Branch 343 not taken.

✗ Branch 344 not taken.

✗ Branch 345 not taken.

✗ Branch 346 not taken.

✗ Branch 347 not taken.

✗ Branch 348 not taken.

✗ Branch 349 not taken.

✗ Branch 350 not taken.

✗ Branch 351 not taken.

✗ Branch 352 not taken.

✗ Branch 353 not taken.

✗ Branch 354 not taken.

✗ Branch 355 not taken.

✗ Branch 356 not taken.

✗ Branch 357 not taken.

✗ Branch 358 not taken.

✗ Branch 359 not taken.

✗ Branch 360 not taken.

✗ Branch 361 not taken.

✗ Branch 362 not taken.

✗ Branch 363 not taken.

✗ Branch 364 not taken.

✗ Branch 365 not taken.

✗ Branch 366 not taken.

✗ Branch 367 not taken.

✗ Branch 368 not taken.

✗ Branch 369 not taken.

✗ Branch 370 not taken.

✗ Branch 371 not taken.

✗ Branch 372 not taken.

✗ Branch 373 not taken.

✗ Branch 374 not taken.

✗ Branch 375 not taken.

✗ Branch 376 not taken.

✗ Branch 377 not taken.

✗ Branch 378 not taken.

✗ Branch 379 not taken.

✗ Branch 380 not taken.

✗ Branch 381 not taken.

✗ Branch 382 not taken.

✗ Branch 383 not taken.

✗ Branch 384 not taken.

✗ Branch 385 not taken.

✗ Branch 386 not taken.

✗ Branch 387 not taken.

✗ Branch 388 not taken.

✗ Branch 389 not taken.

✗ Branch 390 not taken.

✗ Branch 391 not taken.

✗ Branch 392 not taken.

✗ Branch 393 not taken.

✗ Branch 394 not taken.

✗ Branch 395 not taken.

✗ Branch 396 not taken.

✗ Branch 397 not taken.

✗ Branch 398 not taken.

✗ Branch 399 not taken.

✗ Branch 400 not taken.

✗ Branch 401 not taken.

✗ Branch 402 not taken.

✗ Branch 403 not taken.

✗ Branch 404 not taken.

✗ Branch 405 not taken.

✗ Branch 406 not taken.

✗ Branch 407 not taken.

✗ Branch 408 not taken.

✗ Branch 409 not taken.

✗ Branch 410 not taken.

✗ Branch 411 not taken.

✗ Branch 412 not taken.

✗ Branch 413 not taken.

✗ Branch 414 not taken.

✗ Branch 415 not taken.

✗ Branch 416 not taken.

✗ Branch 417 not taken.

✗ Branch 418 not taken.

✗ Branch 419 not taken.

✗ Branch 420 not taken.

✗ Branch 421 not taken.

✗ Branch 422 not taken.

✗ Branch 423 not taken.

✗ Branch 424 not taken.

✗ Branch 425 not taken.

✗ Branch 426 not taken.

✗ Branch 427 not taken.

✗ Branch 428 not taken.

✗ Branch 429 not taken.

✗ Branch 430 not taken.

✗ Branch 431 not taken.

✗ Branch 432 not taken.

✗ Branch 433 not taken.

✗ Branch 434 not taken.

✗ Branch 435 not taken.

✗ Branch 436 not taken.

✗ Branch 437 not taken.

✗ Branch 438 not taken.

✗ Branch 439 not taken.

✗ Branch 440 not taken.

✗ Branch 441 not taken.

✗ Branch 442 not taken.

✗ Branch 443 not taken.

2462

if (node_it2 != node_it1->child_nodes_context.end() && node_it2->ch == wc0) {

1114

6/148

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 1613 times.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 639 times.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 56 times.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 57 times.

✗ Branch 89 not taken.

✓ Branch 90 taken 57 times.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✓ Branch 113 taken 40 times.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

2462

if (uca->version == UCA_V900) {

1115

2422

wbeg = node_it2->weight + MY_UCA_900_CE_SIZE + weight_lv;

1116

2422

wbeg_stride = MY_UCA_900_CE_SIZE;

1117

2422

num_of_ce_left = 7;

1118

} else {

1119

40

wbeg = node_it2->weight + 1;

1120

40

wbeg_stride = MY_UCA_900_CE_SIZE;

1121

}

1122

2462

return node_it2->weight + weight_lv;

1123

}

1124

✗

return nullptr;

1125

}

1126

1127

/****************************************************************/

1128

#define HANGUL_JAMO_MAX_LENGTH 3

1129

/**

1130

Check if a code point is Hangul syllable. Decompose it to jamos

1131

if it is, and return tailored weights.

1132

1133

@param syllable Hangul syllable to be decomposed

1134

@param[out] jamo Corresponding jamos

1135

1136

@return 0 The code point is not Hangul syllable

1137

or cannot be decomposed

1138

others The number of jamos returned

1139

*/

1140

408141570

static int my_decompose_hangul_syllable(my_wc_t syllable, my_wc_t *jamo) {

1141

4/4

✓ Branch 0 taken 398320029 times.

✓ Branch 1 taken 9821541 times.

✓ Branch 2 taken 394125971 times.

✓ Branch 3 taken 4194058 times.

408141570

if (syllable < 0xAC00 || syllable > 0xD7AF) return 0;

1142

4194058

constexpr uint syllable_base = 0xAC00;

1143

4194058

constexpr uint leadingjamo_base = 0x1100;

1144

4194058

constexpr uint voweljamo_base = 0x1161;

1145

4194058

constexpr uint trailingjamo_base = 0x11A7;

1146

4194058

constexpr uint voweljamo_cnt = 21;

1147

4194058

constexpr uint trailingjamo_cnt = 28;

1148

4194058

const uint syllable_index = syllable - syllable_base;

1149

4194058

const uint v_t_combination = voweljamo_cnt * trailingjamo_cnt;

1150

4194058

const uint leadingjamo_index = syllable_index / v_t_combination;

1151

4194058

const uint voweljamo_index =

1152

4194058

(syllable_index % v_t_combination) / trailingjamo_cnt;

1153

4194058

const uint trailingjamo_index = syllable_index % trailingjamo_cnt;

1154

4194058

jamo[0] = leadingjamo_base + leadingjamo_index;

1155

4194058

jamo[1] = voweljamo_base + voweljamo_index;

1156

2/2

✓ Branch 0 taken 4043934 times.

✓ Branch 1 taken 150124 times.

4194058

jamo[2] = trailingjamo_index ? (trailingjamo_base + trailingjamo_index) : 0;

1157

2/2

✓ Branch 0 taken 4043934 times.

✓ Branch 1 taken 150124 times.

4194058

return trailingjamo_index ? 3 : 2;

1158

}

1159

1160

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1161

8388116

void uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::my_put_jamo_weights(

1162

my_wc_t *hangul_jamo, int jamo_cnt) {

1163

2/2

✓ Branch 0 taken 12432050 times.

✓ Branch 1 taken 4194058 times.

33252216

for (int jamoind = 0; jamoind < jamo_cnt; jamoind++) {

1164

24864100

uint16 *implicit_weight = implicit + jamoind * MY_UCA_900_CE_SIZE;

1165

24864100

uint page = hangul_jamo[jamoind] >> 8;

1166

24864100

uint code = hangul_jamo[jamoind] & 0xFF;

1167

24864100

const uint16 *jamo_weight_page = uca->weights[page];

1168

24864100

implicit_weight[0] = UCA900_WEIGHT(jamo_weight_page, 0, code);

1169

24864100

implicit_weight[1] = UCA900_WEIGHT(jamo_weight_page, 1, code);

1170

24864100

implicit_weight[2] = UCA900_WEIGHT(jamo_weight_page, 2, code);

1171

}

1172

8388116

implicit[9] = jamo_cnt;

}

/*

Chinese Han characters are assigned an implicit weight according to the

1177

Unicode Collation Algorithm. But when creating our Chinese collation for

1178

utf8mb4, to implement this language's reorder rule, we give the Han

1179

characters in CLDR zh.xml file weight values from 0x1C47 to 0xBDBE, and let

1180

the other Han characters keep their implicit weight. Per UCA, the smallest

1181

leading primary weight of the implicit weight is 0xFB00, and the largest

1182

primary weight we ocuppy for the Han characters in zh.xml is 0xBDBE. There is

1183

a huge gap between these two weight values. To use this weight value gap and

1184

let the character groups like Latin, Cyrillic, have a single primary weight as

1185

before reordering, we change the leading primary weight of the implicit weight

1186

as below.

1187

*/

1188

10147731

static uint16 change_zh_implicit(uint16 weight) {

1189

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 10147731 times.

10147731

assert(weight >= 0xFB00);

1190

6/7

✓ Branch 0 taken 62409 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 1884 times.

✓ Branch 3 taken 471 times.

✓ Branch 4 taken 10519 times.

✓ Branch 5 taken 51845 times.

✓ Branch 6 taken 10020603 times.

10147731

switch (weight) {

1191

62409

case 0xFB00:

1192

62409

return 0xF621;

1193

✗

case 0xFB40:

1194

✗

return 0xBDBF;

1195

1884

case 0xFB41:

1196

1884

return 0xBDC0;

1197

471

case 0xFB80:

1198

471

return 0xBDC1;

1199

10519

case 0xFB84:

1200

10519

return 0xBDC2;

1201

51845

case 0xFB85:

1202

51845

return 0xBDC3;

1203

10020603

default:

1204

10020603

return weight + 0xF622 - 0xFBC0;

}

}

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1209

ALWAYS_INLINE int uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::next_implicit(

1210

my_wc_t ch) {

1211

my_wc_t hangul_jamo[HANGUL_JAMO_MAX_LENGTH];

1212

int jamo_cnt;

1213

21/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 99072 times.

✓ Branch 17 taken 9377766 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 2972160 times.

✓ Branch 21 taken 286125393 times.

✓ Branch 22 taken 99072 times.

✓ Branch 23 taken 9566649 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 66054 times.

✓ Branch 27 taken 6377946 times.

✓ Branch 28 taken 924672 times.

✓ Branch 29 taken 89284608 times.

✓ Branch 30 taken 33024 times.

✓ Branch 31 taken 3195748 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✓ Branch 63 taken 26 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 3 times.

✓ Branch 95 taken 139 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 119 times.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✓ Branch 101 taken 138 times.

✗ Branch 102 not taken.

✓ Branch 103 taken 117 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 1 times.

✓ Branch 107 taken 8034 times.

✗ Branch 108 not taken.

✓ Branch 109 taken 10829 times.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

408141570

if ((jamo_cnt = my_decompose_hangul_syllable(ch, hangul_jamo))) {

1214

4194058

my_put_jamo_weights(hangul_jamo, jamo_cnt);

1215

4194058

num_of_ce_left = jamo_cnt - 1;

1216

4194058

wbeg = implicit + MY_UCA_900_CE_SIZE + weight_lv;

1217

4194058

wbeg_stride = MY_UCA_900_CE_SIZE;

1218

4194058

return *(implicit + weight_lv);

}

/*

We give the Chinese collation different leading primary weight to make

1223

sure there are enough single weight values to be assigned to character

1224

groups like Latin, Cyrillic, etc.

1225

*/

1226

uint page;

1227

41/224

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✓ Branch 32 taken 9105408 times.

✓ Branch 33 taken 272358 times.

✓ Branch 34 taken 62208 times.

✓ Branch 35 taken 9043200 times.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✓ Branch 40 taken 272736543 times.

✓ Branch 41 taken 13388850 times.

✓ Branch 42 taken 1866441 times.

✓ Branch 43 taken 270870102 times.

✓ Branch 44 taken 9105489 times.

✓ Branch 45 taken 461160 times.

✓ Branch 46 taken 62208 times.

✓ Branch 47 taken 9043281 times.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 6070326 times.

✓ Branch 53 taken 307620 times.

✓ Branch 54 taken 41472 times.

✓ Branch 55 taken 6028854 times.

✓ Branch 56 taken 84983808 times.

✓ Branch 57 taken 4300800 times.

✓ Branch 58 taken 580608 times.

✓ Branch 59 taken 84403200 times.

✓ Branch 60 taken 3041515 times.

✓ Branch 61 taken 154233 times.

✓ Branch 62 taken 20736 times.

✓ Branch 63 taken 3020779 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✓ Branch 124 taken 2 times.

✓ Branch 125 taken 24 times.

✗ Branch 126 not taken.

✓ Branch 127 taken 2 times.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✗ Branch 148 not taken.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✓ Branch 188 taken 139 times.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✓ Branch 191 taken 139 times.

✓ Branch 192 taken 119 times.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✓ Branch 195 taken 119 times.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✓ Branch 200 taken 138 times.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✓ Branch 203 taken 138 times.

✓ Branch 204 taken 117 times.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✓ Branch 207 taken 117 times.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✓ Branch 212 taken 4328 times.

✓ Branch 213 taken 3706 times.

✗ Branch 214 not taken.

✓ Branch 215 taken 4328 times.

✓ Branch 216 taken 7191 times.

✓ Branch 217 taken 3638 times.

✗ Branch 218 not taken.

✓ Branch 219 taken 7191 times.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

403947512

if (ch >= 0x17000 && ch <= 0x18AFF) // Tangut character

1228

{

1229

2633673

page = 0xFB00;

1230

2633673

implicit[3] = (ch - 0x17000) | 0x8000;

1231

} else {

1232

401313839

page = ch >> 15;

1233

401313839

implicit[3] = (ch & 0x7FFF) | 0x8000;

1234

106/560

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✓ Branch 80 taken 9315558 times.

✗ Branch 81 not taken.

✓ Branch 82 taken 9257472 times.

✓ Branch 83 taken 58086 times.

✓ Branch 84 taken 8835840 times.

✓ Branch 85 taken 421632 times.

✓ Branch 86 taken 8451441 times.

✓ Branch 87 taken 384399 times.

✓ Branch 88 taken 8451072 times.

✓ Branch 89 taken 422001 times.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✓ Branch 100 taken 284258952 times.

✗ Branch 101 not taken.

✓ Branch 102 taken 282588246 times.

✓ Branch 103 taken 1670706 times.

✓ Branch 104 taken 264649302 times.

✓ Branch 105 taken 17938944 times.

✓ Branch 106 taken 253501731 times.

✓ Branch 107 taken 11147571 times.

✓ Branch 108 taken 253491030 times.

✓ Branch 109 taken 17949645 times.

✓ Branch 110 taken 9504441 times.

✗ Branch 111 not taken.

✓ Branch 112 taken 9446841 times.

✓ Branch 113 taken 57600 times.

✓ Branch 114 taken 8835897 times.

✓ Branch 115 taken 610944 times.

✓ Branch 116 taken 8451498 times.

✓ Branch 117 taken 384399 times.

✓ Branch 118 taken 8451129 times.

✓ Branch 119 taken 611313 times.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✓ Branch 130 taken 6336474 times.

✗ Branch 131 not taken.

✓ Branch 132 taken 6298074 times.

✓ Branch 133 taken 38400 times.

✓ Branch 134 taken 5890598 times.

✓ Branch 135 taken 407476 times.

✓ Branch 136 taken 5634332 times.

✓ Branch 137 taken 256266 times.

✓ Branch 138 taken 5634086 times.

✓ Branch 139 taken 407722 times.

✓ Branch 140 taken 88704000 times.

✗ Branch 141 not taken.

✓ Branch 142 taken 88166400 times.

✓ Branch 143 taken 537600 times.

✓ Branch 144 taken 82467840 times.

✓ Branch 145 taken 5698560 times.

✓ Branch 146 taken 78880116 times.

✓ Branch 147 taken 3587724 times.

✓ Branch 148 taken 78876672 times.

✓ Branch 149 taken 5702004 times.

✓ Branch 150 taken 3175012 times.

✗ Branch 151 not taken.

✓ Branch 152 taken 3155810 times.

✓ Branch 153 taken 19202 times.

✓ Branch 154 taken 2951651 times.

✓ Branch 155 taken 204159 times.

✓ Branch 156 taken 2823515 times.

✓ Branch 157 taken 128136 times.

✓ Branch 158 taken 2823392 times.

✓ Branch 159 taken 204282 times.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✗ Branch 213 not taken.

✗ Branch 214 not taken.

✗ Branch 215 not taken.

✗ Branch 216 not taken.

✗ Branch 217 not taken.

✗ Branch 218 not taken.

✗ Branch 219 not taken.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

✗ Branch 224 not taken.

✗ Branch 225 not taken.

✗ Branch 226 not taken.

✗ Branch 227 not taken.

✗ Branch 228 not taken.

✗ Branch 229 not taken.

✗ Branch 230 not taken.

✗ Branch 231 not taken.

✗ Branch 232 not taken.

✗ Branch 233 not taken.

✗ Branch 234 not taken.

✗ Branch 235 not taken.

✗ Branch 236 not taken.

✗ Branch 237 not taken.

✗ Branch 238 not taken.

✗ Branch 239 not taken.

✗ Branch 240 not taken.

✗ Branch 241 not taken.

✗ Branch 242 not taken.

✗ Branch 243 not taken.

✗ Branch 244 not taken.

✗ Branch 245 not taken.

✗ Branch 246 not taken.

✗ Branch 247 not taken.

✗ Branch 248 not taken.

✗ Branch 249 not taken.

✗ Branch 250 not taken.

✗ Branch 251 not taken.

✗ Branch 252 not taken.

✗ Branch 253 not taken.

✗ Branch 254 not taken.

✗ Branch 255 not taken.

✗ Branch 256 not taken.

✗ Branch 257 not taken.

✗ Branch 258 not taken.

✗ Branch 259 not taken.

✗ Branch 260 not taken.

✗ Branch 261 not taken.

✗ Branch 262 not taken.

✗ Branch 263 not taken.

✗ Branch 264 not taken.

✗ Branch 265 not taken.

✗ Branch 266 not taken.

✗ Branch 267 not taken.

✗ Branch 268 not taken.

✗ Branch 269 not taken.

✗ Branch 270 not taken.

✗ Branch 271 not taken.

✗ Branch 272 not taken.

✗ Branch 273 not taken.

✗ Branch 274 not taken.

✗ Branch 275 not taken.

✗ Branch 276 not taken.

✗ Branch 277 not taken.

✗ Branch 278 not taken.

✗ Branch 279 not taken.

✗ Branch 280 not taken.

✗ Branch 281 not taken.

✗ Branch 282 not taken.

✗ Branch 283 not taken.

✗ Branch 284 not taken.

✗ Branch 285 not taken.

✗ Branch 286 not taken.

✗ Branch 287 not taken.

✗ Branch 288 not taken.

✗ Branch 289 not taken.

✗ Branch 290 not taken.

✗ Branch 291 not taken.

✗ Branch 292 not taken.

✗ Branch 293 not taken.

✗ Branch 294 not taken.

✗ Branch 295 not taken.

✗ Branch 296 not taken.

✗ Branch 297 not taken.

✗ Branch 298 not taken.

✗ Branch 299 not taken.

✗ Branch 300 not taken.

✗ Branch 301 not taken.

✗ Branch 302 not taken.

✗ Branch 303 not taken.

✗ Branch 304 not taken.

✗ Branch 305 not taken.

✗ Branch 306 not taken.

✗ Branch 307 not taken.

✗ Branch 308 not taken.

✗ Branch 309 not taken.

✓ Branch 310 taken 26 times.

✗ Branch 311 not taken.

✓ Branch 312 taken 26 times.

✗ Branch 313 not taken.

✓ Branch 314 taken 2 times.

✓ Branch 315 taken 24 times.

✗ Branch 316 not taken.

✓ Branch 317 taken 2 times.

✗ Branch 318 not taken.

✓ Branch 319 taken 24 times.

✗ Branch 320 not taken.

✗ Branch 321 not taken.

✗ Branch 322 not taken.

✗ Branch 323 not taken.

✗ Branch 324 not taken.

✗ Branch 325 not taken.

✗ Branch 326 not taken.

✗ Branch 327 not taken.

✗ Branch 328 not taken.

✗ Branch 329 not taken.

✗ Branch 330 not taken.

✗ Branch 331 not taken.

✗ Branch 332 not taken.

✗ Branch 333 not taken.

✗ Branch 334 not taken.

✗ Branch 335 not taken.

✗ Branch 336 not taken.

✗ Branch 337 not taken.

✗ Branch 338 not taken.

✗ Branch 339 not taken.

✗ Branch 340 not taken.

✗ Branch 341 not taken.

✗ Branch 342 not taken.

✗ Branch 343 not taken.

✗ Branch 344 not taken.

✗ Branch 345 not taken.

✗ Branch 346 not taken.

✗ Branch 347 not taken.

✗ Branch 348 not taken.

✗ Branch 349 not taken.

✗ Branch 350 not taken.

✗ Branch 351 not taken.

✗ Branch 352 not taken.

✗ Branch 353 not taken.

✗ Branch 354 not taken.

✗ Branch 355 not taken.

✗ Branch 356 not taken.

✗ Branch 357 not taken.

✗ Branch 358 not taken.

✗ Branch 359 not taken.

✗ Branch 360 not taken.

✗ Branch 361 not taken.

✗ Branch 362 not taken.

✗ Branch 363 not taken.

✗ Branch 364 not taken.

✗ Branch 365 not taken.

✗ Branch 366 not taken.

✗ Branch 367 not taken.

✗ Branch 368 not taken.

✗ Branch 369 not taken.

✗ Branch 370 not taken.

✗ Branch 371 not taken.

✗ Branch 372 not taken.

✗ Branch 373 not taken.

✗ Branch 374 not taken.

✗ Branch 375 not taken.

✗ Branch 376 not taken.

✗ Branch 377 not taken.

✗ Branch 378 not taken.

✗ Branch 379 not taken.

✗ Branch 380 not taken.

✗ Branch 381 not taken.

✗ Branch 382 not taken.

✗ Branch 383 not taken.

✗ Branch 384 not taken.

✗ Branch 385 not taken.

✗ Branch 386 not taken.

✗ Branch 387 not taken.

✗ Branch 388 not taken.

✗ Branch 389 not taken.

✗ Branch 390 not taken.

✗ Branch 391 not taken.

✗ Branch 392 not taken.

✗ Branch 393 not taken.

✗ Branch 394 not taken.

✗ Branch 395 not taken.

✗ Branch 396 not taken.

✗ Branch 397 not taken.

✗ Branch 398 not taken.

✗ Branch 399 not taken.

✗ Branch 400 not taken.

✗ Branch 401 not taken.

✗ Branch 402 not taken.

✗ Branch 403 not taken.

✗ Branch 404 not taken.

✗ Branch 405 not taken.

✗ Branch 406 not taken.

✗ Branch 407 not taken.

✗ Branch 408 not taken.

✗ Branch 409 not taken.

✗ Branch 410 not taken.

✗ Branch 411 not taken.

✗ Branch 412 not taken.

✗ Branch 413 not taken.

✗ Branch 414 not taken.

✗ Branch 415 not taken.

✗ Branch 416 not taken.

✗ Branch 417 not taken.

✗ Branch 418 not taken.

✗ Branch 419 not taken.

✗ Branch 420 not taken.

✗ Branch 421 not taken.

✗ Branch 422 not taken.

✗ Branch 423 not taken.

✗ Branch 424 not taken.

✗ Branch 425 not taken.

✗ Branch 426 not taken.

✗ Branch 427 not taken.

✗ Branch 428 not taken.

✗ Branch 429 not taken.

✗ Branch 430 not taken.

✗ Branch 431 not taken.

✗ Branch 432 not taken.

✗ Branch 433 not taken.

✗ Branch 434 not taken.

✗ Branch 435 not taken.

✗ Branch 436 not taken.

✗ Branch 437 not taken.

✗ Branch 438 not taken.

✗ Branch 439 not taken.

✗ Branch 440 not taken.

✗ Branch 441 not taken.

✗ Branch 442 not taken.

✗ Branch 443 not taken.

✗ Branch 444 not taken.

✗ Branch 445 not taken.

✗ Branch 446 not taken.

✗ Branch 447 not taken.

✗ Branch 448 not taken.

✗ Branch 449 not taken.

✗ Branch 450 not taken.

✗ Branch 451 not taken.

✗ Branch 452 not taken.

✗ Branch 453 not taken.

✗ Branch 454 not taken.

✗ Branch 455 not taken.

✗ Branch 456 not taken.

✗ Branch 457 not taken.

✗ Branch 458 not taken.

✗ Branch 459 not taken.

✗ Branch 460 not taken.

✗ Branch 461 not taken.

✗ Branch 462 not taken.

✗ Branch 463 not taken.

✗ Branch 464 not taken.

✗ Branch 465 not taken.

✗ Branch 466 not taken.

✗ Branch 467 not taken.

✗ Branch 468 not taken.

✗ Branch 469 not taken.

✓ Branch 470 taken 139 times.

✗ Branch 471 not taken.

✓ Branch 472 taken 139 times.

✗ Branch 473 not taken.

✓ Branch 474 taken 100 times.

✓ Branch 475 taken 39 times.

✓ Branch 476 taken 100 times.

✗ Branch 477 not taken.

✓ Branch 478 taken 100 times.

✓ Branch 479 taken 39 times.

✓ Branch 480 taken 119 times.

✗ Branch 481 not taken.

✓ Branch 482 taken 119 times.

✗ Branch 483 not taken.

✓ Branch 484 taken 94 times.

✓ Branch 485 taken 25 times.

✓ Branch 486 taken 94 times.

✗ Branch 487 not taken.

✓ Branch 488 taken 94 times.

✓ Branch 489 taken 25 times.

✗ Branch 490 not taken.

✗ Branch 491 not taken.

✗ Branch 492 not taken.

✗ Branch 493 not taken.

✗ Branch 494 not taken.

✗ Branch 495 not taken.

✗ Branch 496 not taken.

✗ Branch 497 not taken.

✗ Branch 498 not taken.

✗ Branch 499 not taken.

✓ Branch 500 taken 138 times.

✗ Branch 501 not taken.

✓ Branch 502 taken 138 times.

✗ Branch 503 not taken.

✓ Branch 504 taken 99 times.

✓ Branch 505 taken 39 times.

✓ Branch 506 taken 99 times.

✗ Branch 507 not taken.

✓ Branch 508 taken 99 times.

✓ Branch 509 taken 39 times.

✓ Branch 510 taken 117 times.

✗ Branch 511 not taken.

✓ Branch 512 taken 117 times.

✗ Branch 513 not taken.

✓ Branch 514 taken 93 times.

✓ Branch 515 taken 24 times.

✓ Branch 516 taken 93 times.

✗ Branch 517 not taken.

✓ Branch 518 taken 93 times.

✓ Branch 519 taken 24 times.

✗ Branch 520 not taken.

✗ Branch 521 not taken.

✗ Branch 522 not taken.

✗ Branch 523 not taken.

✗ Branch 524 not taken.

✗ Branch 525 not taken.

✗ Branch 526 not taken.

✗ Branch 527 not taken.

✗ Branch 528 not taken.

✗ Branch 529 not taken.

✓ Branch 530 taken 8034 times.

✗ Branch 531 not taken.

✓ Branch 532 taken 7785 times.

✓ Branch 533 taken 249 times.

✓ Branch 534 taken 4289 times.

✓ Branch 535 taken 3496 times.

✓ Branch 536 taken 3637 times.

✓ Branch 537 taken 652 times.

✓ Branch 538 taken 3637 times.

✓ Branch 539 taken 3496 times.

✓ Branch 540 taken 10829 times.

✗ Branch 541 not taken.

✓ Branch 542 taken 10631 times.

✓ Branch 543 taken 198 times.

✓ Branch 544 taken 7167 times.

✓ Branch 545 taken 3464 times.

✓ Branch 546 taken 6541 times.

✓ Branch 547 taken 626 times.

✓ Branch 548 taken 6541 times.

✓ Branch 549 taken 3464 times.

✗ Branch 550 not taken.

✗ Branch 551 not taken.

✗ Branch 552 not taken.

✗ Branch 553 not taken.

✗ Branch 554 not taken.

✗ Branch 555 not taken.

✗ Branch 556 not taken.

✗ Branch 557 not taken.

✗ Branch 558 not taken.

✗ Branch 559 not taken.

401313839

if ((ch >= 0x3400 && ch <= 0x4DB5) || (ch >= 0x20000 && ch <= 0x2A6D6) ||

1235

92/448

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✓ Branch 64 taken 8413731 times.

✓ Branch 65 taken 37341 times.

✓ Branch 66 taken 8413632 times.

✓ Branch 67 taken 422100 times.

✓ Branch 68 taken 8411634 times.

✓ Branch 69 taken 1998 times.

✓ Branch 70 taken 8411616 times.

✓ Branch 71 taken 422118 times.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✓ Branch 80 taken 252398448 times.

✓ Branch 81 taken 1092582 times.

✓ Branch 82 taken 252395478 times.

✓ Branch 83 taken 17952615 times.

✓ Branch 84 taken 252335808 times.

✓ Branch 85 taken 59670 times.

✓ Branch 86 taken 252335286 times.

✓ Branch 87 taken 17953137 times.

✓ Branch 88 taken 8413788 times.

✓ Branch 89 taken 37341 times.

✓ Branch 90 taken 8413689 times.

✓ Branch 91 taken 611412 times.

✓ Branch 92 taken 8411667 times.

✓ Branch 93 taken 2022 times.

✓ Branch 94 taken 8411649 times.

✓ Branch 95 taken 611430 times.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✓ Branch 104 taken 5609192 times.

✓ Branch 105 taken 24894 times.

✓ Branch 106 taken 5609126 times.

✓ Branch 107 taken 407788 times.

✓ Branch 108 taken 5607778 times.

✓ Branch 109 taken 1348 times.

✓ Branch 110 taken 5607766 times.

✓ Branch 111 taken 407800 times.

✓ Branch 112 taken 78528156 times.

✓ Branch 113 taken 348516 times.

✓ Branch 114 taken 78527232 times.

✓ Branch 115 taken 5702928 times.

✓ Branch 116 taken 78508584 times.

✓ Branch 117 taken 18648 times.

✓ Branch 118 taken 78508416 times.

✓ Branch 119 taken 5703096 times.

✓ Branch 120 taken 2810945 times.

✓ Branch 121 taken 12447 times.

✓ Branch 122 taken 2810912 times.

✓ Branch 123 taken 204315 times.

✓ Branch 124 taken 2810238 times.

✓ Branch 125 taken 674 times.

✓ Branch 126 taken 2810232 times.

✓ Branch 127 taken 204321 times.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✗ Branch 148 not taken.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✗ Branch 213 not taken.

✗ Branch 214 not taken.

✗ Branch 215 not taken.

✗ Branch 216 not taken.

✗ Branch 217 not taken.

✗ Branch 218 not taken.

✗ Branch 219 not taken.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

✗ Branch 224 not taken.

✗ Branch 225 not taken.

✗ Branch 226 not taken.

✗ Branch 227 not taken.

✗ Branch 228 not taken.

✗ Branch 229 not taken.

✗ Branch 230 not taken.

✗ Branch 231 not taken.

✗ Branch 232 not taken.

✗ Branch 233 not taken.

✗ Branch 234 not taken.

✗ Branch 235 not taken.

✗ Branch 236 not taken.

✗ Branch 237 not taken.

✗ Branch 238 not taken.

✗ Branch 239 not taken.

✗ Branch 240 not taken.

✗ Branch 241 not taken.

✗ Branch 242 not taken.

✗ Branch 243 not taken.

✗ Branch 244 not taken.

✗ Branch 245 not taken.

✗ Branch 246 not taken.

✗ Branch 247 not taken.

✗ Branch 248 not taken.

✗ Branch 249 not taken.

✗ Branch 250 not taken.

✓ Branch 251 taken 24 times.

✗ Branch 252 not taken.

✗ Branch 253 not taken.

✗ Branch 254 not taken.

✓ Branch 255 taken 24 times.

✗ Branch 256 not taken.

✗ Branch 257 not taken.

✗ Branch 258 not taken.

✗ Branch 259 not taken.

✗ Branch 260 not taken.

✗ Branch 261 not taken.

✗ Branch 262 not taken.

✗ Branch 263 not taken.

✗ Branch 264 not taken.

✗ Branch 265 not taken.

✗ Branch 266 not taken.

✗ Branch 267 not taken.

✗ Branch 268 not taken.

✗ Branch 269 not taken.

✗ Branch 270 not taken.

✗ Branch 271 not taken.

✗ Branch 272 not taken.

✗ Branch 273 not taken.

✗ Branch 274 not taken.

✗ Branch 275 not taken.

✗ Branch 276 not taken.

✗ Branch 277 not taken.

✗ Branch 278 not taken.

✗ Branch 279 not taken.

✗ Branch 280 not taken.

✗ Branch 281 not taken.

✗ Branch 282 not taken.

✗ Branch 283 not taken.

✗ Branch 284 not taken.

✗ Branch 285 not taken.

✗ Branch 286 not taken.

✗ Branch 287 not taken.

✗ Branch 288 not taken.

✗ Branch 289 not taken.

✗ Branch 290 not taken.

✗ Branch 291 not taken.

✗ Branch 292 not taken.

✗ Branch 293 not taken.

✗ Branch 294 not taken.

✗ Branch 295 not taken.

✗ Branch 296 not taken.

✗ Branch 297 not taken.

✗ Branch 298 not taken.

✗ Branch 299 not taken.

✗ Branch 300 not taken.

✗ Branch 301 not taken.

✗ Branch 302 not taken.

✗ Branch 303 not taken.

✗ Branch 304 not taken.

✗ Branch 305 not taken.

✗ Branch 306 not taken.

✗ Branch 307 not taken.

✗ Branch 308 not taken.

✗ Branch 309 not taken.

✗ Branch 310 not taken.

✗ Branch 311 not taken.

✗ Branch 312 not taken.

✗ Branch 313 not taken.

✗ Branch 314 not taken.

✗ Branch 315 not taken.

✗ Branch 316 not taken.

✗ Branch 317 not taken.

✗ Branch 318 not taken.

✗ Branch 319 not taken.

✗ Branch 320 not taken.

✗ Branch 321 not taken.

✗ Branch 322 not taken.

✗ Branch 323 not taken.

✗ Branch 324 not taken.

✗ Branch 325 not taken.

✗ Branch 326 not taken.

✗ Branch 327 not taken.

✗ Branch 328 not taken.

✗ Branch 329 not taken.

✗ Branch 330 not taken.

✗ Branch 331 not taken.

✗ Branch 332 not taken.

✗ Branch 333 not taken.

✗ Branch 334 not taken.

✗ Branch 335 not taken.

✗ Branch 336 not taken.

✗ Branch 337 not taken.

✗ Branch 338 not taken.

✗ Branch 339 not taken.

✗ Branch 340 not taken.

✗ Branch 341 not taken.

✗ Branch 342 not taken.

✗ Branch 343 not taken.

✗ Branch 344 not taken.

✗ Branch 345 not taken.

✗ Branch 346 not taken.

✗ Branch 347 not taken.

✗ Branch 348 not taken.

✗ Branch 349 not taken.

✗ Branch 350 not taken.

✗ Branch 351 not taken.

✗ Branch 352 not taken.

✗ Branch 353 not taken.

✗ Branch 354 not taken.

✗ Branch 355 not taken.

✗ Branch 356 not taken.

✗ Branch 357 not taken.

✗ Branch 358 not taken.

✗ Branch 359 not taken.

✗ Branch 360 not taken.

✗ Branch 361 not taken.

✗ Branch 362 not taken.

✗ Branch 363 not taken.

✗ Branch 364 not taken.

✗ Branch 365 not taken.

✗ Branch 366 not taken.

✗ Branch 367 not taken.

✗ Branch 368 not taken.

✗ Branch 369 not taken.

✗ Branch 370 not taken.

✗ Branch 371 not taken.

✗ Branch 372 not taken.

✗ Branch 373 not taken.

✗ Branch 374 not taken.

✗ Branch 375 not taken.

✓ Branch 376 taken 100 times.

✗ Branch 377 not taken.

✓ Branch 378 taken 100 times.

✓ Branch 379 taken 39 times.

✓ Branch 380 taken 56 times.

✓ Branch 381 taken 44 times.

✓ Branch 382 taken 56 times.

✓ Branch 383 taken 39 times.

✓ Branch 384 taken 94 times.

✗ Branch 385 not taken.

✓ Branch 386 taken 94 times.

✓ Branch 387 taken 25 times.

✓ Branch 388 taken 61 times.

✓ Branch 389 taken 33 times.

✓ Branch 390 taken 61 times.

✓ Branch 391 taken 25 times.

✗ Branch 392 not taken.

✗ Branch 393 not taken.

✗ Branch 394 not taken.

✗ Branch 395 not taken.

✗ Branch 396 not taken.

✗ Branch 397 not taken.

✗ Branch 398 not taken.

✗ Branch 399 not taken.

✓ Branch 400 taken 99 times.

✗ Branch 401 not taken.

✓ Branch 402 taken 99 times.

✓ Branch 403 taken 39 times.

✓ Branch 404 taken 55 times.

✓ Branch 405 taken 44 times.

✓ Branch 406 taken 55 times.

✓ Branch 407 taken 39 times.

✓ Branch 408 taken 93 times.

✗ Branch 409 not taken.

✓ Branch 410 taken 93 times.

✓ Branch 411 taken 24 times.

✓ Branch 412 taken 60 times.

✓ Branch 413 taken 33 times.

✓ Branch 414 taken 60 times.

✓ Branch 415 taken 24 times.

✗ Branch 416 not taken.

✗ Branch 417 not taken.

✗ Branch 418 not taken.

✗ Branch 419 not taken.

✗ Branch 420 not taken.

✗ Branch 421 not taken.

✗ Branch 422 not taken.

✗ Branch 423 not taken.

✓ Branch 424 taken 3637 times.

✗ Branch 425 not taken.

✓ Branch 426 taken 3637 times.

✓ Branch 427 taken 3496 times.

✓ Branch 428 taken 3593 times.

✓ Branch 429 taken 44 times.

✓ Branch 430 taken 3593 times.

✓ Branch 431 taken 3496 times.

✓ Branch 432 taken 6541 times.

✗ Branch 433 not taken.

✓ Branch 434 taken 6541 times.

✓ Branch 435 taken 3464 times.

✓ Branch 436 taken 6508 times.

✓ Branch 437 taken 33 times.

✓ Branch 438 taken 6508 times.

✓ Branch 439 taken 3464 times.

✗ Branch 440 not taken.

✗ Branch 441 not taken.

✗ Branch 442 not taken.

✗ Branch 443 not taken.

✗ Branch 444 not taken.

✗ Branch 445 not taken.

✗ Branch 446 not taken.

✗ Branch 447 not taken.

383042023

(ch >= 0x2A700 && ch <= 0x2B734) || (ch >= 0x2B740 && ch <= 0x2B81D) ||

1236

24/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 51858 times.

✓ Branch 17 taken 8359758 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 1543050 times.

✓ Branch 21 taken 250792236 times.

✓ Branch 22 taken 51882 times.

✓ Branch 23 taken 8359767 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 34588 times.

✓ Branch 27 taken 5573178 times.

✓ Branch 28 taken 484008 times.

✓ Branch 29 taken 78024408 times.

✓ Branch 30 taken 17294 times.

✓ Branch 31 taken 2792938 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 31 times.

✓ Branch 95 taken 25 times.

✓ Branch 96 taken 44 times.

✓ Branch 97 taken 17 times.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✓ Branch 100 taken 31 times.

✓ Branch 101 taken 24 times.

✓ Branch 102 taken 44 times.

✓ Branch 103 taken 16 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 31 times.

✓ Branch 107 taken 3562 times.

✓ Branch 108 taken 44 times.

✓ Branch 109 taken 6464 times.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

356095298

(ch >= 0x2B820 && ch <= 0x2CEA1)) {

1237

22092433

page += 0xFB80;

1238

63/448

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✓ Branch 64 taken 8781876 times.

✗ Branch 65 not taken.

✓ Branch 66 taken 8781876 times.

✗ Branch 67 not taken.

✓ Branch 68 taken 8724276 times.

✓ Branch 69 taken 57600 times.

✗ Branch 70 not taken.

✓ Branch 71 taken 8724276 times.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✓ Branch 80 taken 268745373 times.

✗ Branch 81 not taken.

✓ Branch 82 taken 263465973 times.

✓ Branch 83 taken 5279400 times.

✓ Branch 84 taken 261727389 times.

✓ Branch 85 taken 1738584 times.

✗ Branch 86 not taken.

✓ Branch 87 taken 261727389 times.

✓ Branch 88 taken 8971197 times.

✗ Branch 89 not taken.

✓ Branch 90 taken 8782287 times.

✓ Branch 91 taken 188910 times.

✓ Branch 92 taken 8724309 times.

✓ Branch 93 taken 57978 times.

✗ Branch 94 not taken.

✓ Branch 95 taken 8724309 times.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✓ Branch 104 taken 5980978 times.

✗ Branch 105 not taken.

✓ Branch 106 taken 5854858 times.

✓ Branch 107 taken 126120 times.

✓ Branch 108 taken 5816206 times.

✓ Branch 109 taken 38652 times.

✗ Branch 110 not taken.

✓ Branch 111 taken 5816206 times.

✓ Branch 112 taken 83727504 times.

✗ Branch 113 not taken.

✓ Branch 114 taken 81967704 times.

✓ Branch 115 taken 1759800 times.

✓ Branch 116 taken 81426576 times.

✓ Branch 117 taken 541128 times.

✗ Branch 118 not taken.

✓ Branch 119 taken 81426576 times.

✓ Branch 120 taken 2997259 times.

✗ Branch 121 not taken.

✓ Branch 122 taken 2933778 times.

✓ Branch 123 taken 63481 times.

✓ Branch 124 taken 2914452 times.

✓ Branch 125 taken 19326 times.

✗ Branch 126 not taken.

✓ Branch 127 taken 2914452 times.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✗ Branch 148 not taken.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✗ Branch 213 not taken.

✗ Branch 214 not taken.

✗ Branch 215 not taken.

✗ Branch 216 not taken.

✗ Branch 217 not taken.

✗ Branch 218 not taken.

✗ Branch 219 not taken.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

✗ Branch 224 not taken.

✗ Branch 225 not taken.

✗ Branch 226 not taken.

✗ Branch 227 not taken.

✗ Branch 228 not taken.

✗ Branch 229 not taken.

✗ Branch 230 not taken.

✗ Branch 231 not taken.

✗ Branch 232 not taken.

✗ Branch 233 not taken.

✗ Branch 234 not taken.

✗ Branch 235 not taken.

✗ Branch 236 not taken.

✗ Branch 237 not taken.

✗ Branch 238 not taken.

✗ Branch 239 not taken.

✗ Branch 240 not taken.

✗ Branch 241 not taken.

✗ Branch 242 not taken.

✗ Branch 243 not taken.

✗ Branch 244 not taken.

✗ Branch 245 not taken.

✗ Branch 246 not taken.

✗ Branch 247 not taken.

✓ Branch 248 taken 24 times.

✗ Branch 249 not taken.

✗ Branch 250 not taken.

✓ Branch 251 taken 24 times.

✗ Branch 252 not taken.

✗ Branch 253 not taken.

✗ Branch 254 not taken.

✗ Branch 255 not taken.

✗ Branch 256 not taken.

✗ Branch 257 not taken.

✗ Branch 258 not taken.

✗ Branch 259 not taken.

✗ Branch 260 not taken.

✗ Branch 261 not taken.

✗ Branch 262 not taken.

✗ Branch 263 not taken.

✗ Branch 264 not taken.

✗ Branch 265 not taken.

✗ Branch 266 not taken.

✗ Branch 267 not taken.

✗ Branch 268 not taken.

✗ Branch 269 not taken.

✗ Branch 270 not taken.

✗ Branch 271 not taken.

✗ Branch 272 not taken.

✗ Branch 273 not taken.

✗ Branch 274 not taken.

✗ Branch 275 not taken.

✗ Branch 276 not taken.

✗ Branch 277 not taken.

✗ Branch 278 not taken.

✗ Branch 279 not taken.

✗ Branch 280 not taken.

✗ Branch 281 not taken.

✗ Branch 282 not taken.

✗ Branch 283 not taken.

✗ Branch 284 not taken.

✗ Branch 285 not taken.

✗ Branch 286 not taken.

✗ Branch 287 not taken.

✗ Branch 288 not taken.

✗ Branch 289 not taken.

✗ Branch 290 not taken.

✗ Branch 291 not taken.

✗ Branch 292 not taken.

✗ Branch 293 not taken.

✗ Branch 294 not taken.

✗ Branch 295 not taken.

✗ Branch 296 not taken.

✗ Branch 297 not taken.

✗ Branch 298 not taken.

✗ Branch 299 not taken.

✗ Branch 300 not taken.

✗ Branch 301 not taken.

✗ Branch 302 not taken.

✗ Branch 303 not taken.

✗ Branch 304 not taken.

✗ Branch 305 not taken.

✗ Branch 306 not taken.

✗ Branch 307 not taken.

✗ Branch 308 not taken.

✗ Branch 309 not taken.

✗ Branch 310 not taken.

✗ Branch 311 not taken.

✗ Branch 312 not taken.

✗ Branch 313 not taken.

✗ Branch 314 not taken.

✗ Branch 315 not taken.

✗ Branch 316 not taken.

✗ Branch 317 not taken.

✗ Branch 318 not taken.

✗ Branch 319 not taken.

✗ Branch 320 not taken.

✗ Branch 321 not taken.

✗ Branch 322 not taken.

✗ Branch 323 not taken.

✗ Branch 324 not taken.

✗ Branch 325 not taken.

✗ Branch 326 not taken.

✗ Branch 327 not taken.

✗ Branch 328 not taken.

✗ Branch 329 not taken.

✗ Branch 330 not taken.

✗ Branch 331 not taken.

✗ Branch 332 not taken.

✗ Branch 333 not taken.

✗ Branch 334 not taken.

✗ Branch 335 not taken.

✗ Branch 336 not taken.

✗ Branch 337 not taken.

✗ Branch 338 not taken.

✗ Branch 339 not taken.

✗ Branch 340 not taken.

✗ Branch 341 not taken.

✗ Branch 342 not taken.

✗ Branch 343 not taken.

✗ Branch 344 not taken.

✗ Branch 345 not taken.

✗ Branch 346 not taken.

✗ Branch 347 not taken.

✗ Branch 348 not taken.

✗ Branch 349 not taken.

✗ Branch 350 not taken.

✗ Branch 351 not taken.

✗ Branch 352 not taken.

✗ Branch 353 not taken.

✗ Branch 354 not taken.

✗ Branch 355 not taken.

✗ Branch 356 not taken.

✗ Branch 357 not taken.

✗ Branch 358 not taken.

✗ Branch 359 not taken.

✗ Branch 360 not taken.

✗ Branch 361 not taken.

✗ Branch 362 not taken.

✗ Branch 363 not taken.

✗ Branch 364 not taken.

✗ Branch 365 not taken.

✗ Branch 366 not taken.

✗ Branch 367 not taken.

✗ Branch 368 not taken.

✗ Branch 369 not taken.

✗ Branch 370 not taken.

✗ Branch 371 not taken.

✗ Branch 372 not taken.

✗ Branch 373 not taken.

✗ Branch 374 not taken.

✗ Branch 375 not taken.

✓ Branch 376 taken 64 times.

✗ Branch 377 not taken.

✓ Branch 378 taken 64 times.

✗ Branch 379 not taken.

✓ Branch 380 taken 64 times.

✗ Branch 381 not taken.

✗ Branch 382 not taken.

✓ Branch 383 taken 64 times.

✓ Branch 384 taken 42 times.

✗ Branch 385 not taken.

✓ Branch 386 taken 42 times.

✗ Branch 387 not taken.

✓ Branch 388 taken 42 times.

✗ Branch 389 not taken.

✗ Branch 390 not taken.

✓ Branch 391 taken 42 times.

✗ Branch 392 not taken.

✗ Branch 393 not taken.

✗ Branch 394 not taken.

✗ Branch 395 not taken.

✗ Branch 396 not taken.

✗ Branch 397 not taken.

✗ Branch 398 not taken.

✗ Branch 399 not taken.

✓ Branch 400 taken 63 times.

✗ Branch 401 not taken.

✓ Branch 402 taken 63 times.

✗ Branch 403 not taken.

✓ Branch 404 taken 63 times.

✗ Branch 405 not taken.

✗ Branch 406 not taken.

✓ Branch 407 taken 63 times.

✓ Branch 408 taken 40 times.

✗ Branch 409 not taken.

✓ Branch 410 taken 40 times.

✗ Branch 411 not taken.

✓ Branch 412 taken 40 times.

✗ Branch 413 not taken.

✗ Branch 414 not taken.

✓ Branch 415 taken 40 times.

✗ Branch 416 not taken.

✗ Branch 417 not taken.

✗ Branch 418 not taken.

✗ Branch 419 not taken.

✗ Branch 420 not taken.

✗ Branch 421 not taken.

✗ Branch 422 not taken.

✗ Branch 423 not taken.

✓ Branch 424 taken 7058 times.

✗ Branch 425 not taken.

✓ Branch 426 taken 3601 times.

✓ Branch 427 taken 3457 times.

✓ Branch 428 taken 3601 times.

✗ Branch 429 not taken.

✗ Branch 430 not taken.

✓ Branch 431 taken 3601 times.

✓ Branch 432 taken 9928 times.

✗ Branch 433 not taken.

✓ Branch 434 taken 6488 times.

✓ Branch 435 taken 3440 times.

✓ Branch 436 taken 6488 times.

✗ Branch 437 not taken.

✗ Branch 438 not taken.

✓ Branch 439 taken 6488 times.

✗ Branch 440 not taken.

✗ Branch 441 not taken.

✗ Branch 442 not taken.

✗ Branch 443 not taken.

✗ Branch 444 not taken.

✗ Branch 445 not taken.

✗ Branch 446 not taken.

✗ Branch 447 not taken.

379221406

} else if ((ch >= 0x4E00 && ch <= 0x9FD5) || (ch >= 0xFA0E && ch <= 0xFA29))

1239

7424632

page += 0xFB40;

1240

else

1241

371796774

page += 0xFBC0;

1242

}

1243

14/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✓ Branch 17 taken 9377766 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 8893983 times.

✓ Branch 21 taken 277231410 times.

✗ Branch 22 not taken.

✓ Branch 23 taken 9566649 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 6377946 times.

✗ Branch 28 not taken.

✓ Branch 29 taken 89284608 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 3195748 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✓ Branch 63 taken 26 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✓ Branch 95 taken 139 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 119 times.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✓ Branch 101 taken 138 times.

✗ Branch 102 not taken.

✓ Branch 103 taken 117 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✓ Branch 107 taken 8034 times.

✗ Branch 108 not taken.

✓ Branch 109 taken 10829 times.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

403947512

if (cs->coll_param == &zh_coll_param) {

1244

8893983

page = change_zh_implicit(page);

1245

}

1246

403947512

implicit[0] = page;

1247

403947512

implicit[1] = 0x0020;

1248

403947512

implicit[2] = 0x0002;

1249

// implicit[3] is set above.

1250

403947512

implicit[4] = 0;

1251

403947512

implicit[5] = 0;

1252

403947512

num_of_ce_left = 1;

1253

403947512

wbeg = implicit + MY_UCA_900_CE_SIZE + weight_lv;

1254

403947512

wbeg_stride = MY_UCA_900_CE_SIZE;

1255

1256

403947512

return *(implicit + weight_lv);

}

/**

Return implicit UCA weight

1261

Used for code points that do not have assigned UCA weights.

1262

1263

@return The leading implicit weight.

1264

*/

1265

1266

template <class Mb_wc>

1267

ALWAYS_INLINE int uca_scanner_any<Mb_wc>::next_implicit(my_wc_t ch) {

1268

30495286

implicit[0] = (ch & 0x7FFF) | 0x8000;

1269

30495286

implicit[1] = 0;

1270

30495286

wbeg = implicit;

1271

30495286

wbeg_stride = MY_UCA_900_CE_SIZE;

1272

1273

30495286

uint page = ch >> 15;

1274

1275

10/36

✓ Branch 0 taken 1862400 times.

✓ Branch 1 taken 21439551 times.

✓ Branch 2 taken 460800 times.

✓ Branch 3 taken 6022659 times.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✓ Branch 9 taken 372 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 357 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 12800 times.

✓ Branch 17 taken 76815 times.

✓ Branch 18 taken 12800 times.

✓ Branch 19 taken 76812 times.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

29965366

if (ch >= 0x3400 && ch <= 0x4DB5)

1276

2348800

page += 0xFB80;

1277

18/72

✓ Branch 0 taken 21439551 times.

✓ Branch 1 taken 423936 times.

✓ Branch 2 taken 6137832 times.

✓ Branch 3 taken 15301719 times.

✓ Branch 4 taken 6022659 times.

✓ Branch 5 taken 105984 times.

✓ Branch 6 taken 1504944 times.

✓ Branch 7 taken 4517715 times.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 372 times.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✓ Branch 19 taken 372 times.

✓ Branch 20 taken 357 times.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✓ Branch 23 taken 357 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✓ Branch 32 taken 76815 times.

✗ Branch 33 not taken.

✓ Branch 34 taken 41804 times.

✓ Branch 35 taken 35011 times.

✓ Branch 36 taken 76812 times.

✗ Branch 37 not taken.

✓ Branch 38 taken 41804 times.

✓ Branch 39 taken 35008 times.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

28146486

else if (ch >= 0x4E00 && ch <= 0x9FA5)

1278

7726384

page += 0xFB40;

1279

else

1280

20420102

page += 0xFBC0;

1281

1282

30495286

return page;

1283

}

1284

1285

template <class Mb_wc>

1286

ALWAYS_INLINE int uca_scanner_any<Mb_wc>::next() {

1287

/*

1288

Check if the weights for the previous code point have been

1289

already fully scanned. If yes, then get the next code point and

1290

initialize wbeg and wlength to its weight string.

1291

*/

1292

1293

16/28

✓ Branch 0 taken 25421620 times.

✓ Branch 1 taken 347255757 times.

✓ Branch 2 taken 7002842 times.

✓ Branch 3 taken 160781076 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 102026 times.

✗ Branch 6 not taken.

✓ Branch 7 taken 26946 times.

✓ Branch 8 taken 4312 times.

✓ Branch 9 taken 7436495 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 368 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 1428 times.

✓ Branch 14 taken 101974 times.

✓ Branch 15 taken 290308 times.

✗ Branch 16 not taken.

✓ Branch 17 taken 8 times.

✗ Branch 18 not taken.

✓ Branch 19 taken 20 times.

✗ Branch 20 not taken.

✓ Branch 21 taken 98 times.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✓ Branch 25 taken 84 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

548425362

if (wbeg[0]) /* More weights left from the previous step: */

1294

32635856

return *wbeg++; /* return the next weight from expansion */

1295

1296

do {

1297

523830696

my_wc_t wc = 0;

1298

1299

/* Get next code point */

1300

23/36

✓ Branch 0 taken 347425992 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 80229898 times.

✓ Branch 3 taken 80586689 times.

✓ Branch 4 taken 102026 times.

✗ Branch 5 not taken.

✓ Branch 6 taken 648 times.

✓ Branch 7 taken 26298 times.

✓ Branch 8 taken 7421784 times.

✗ Branch 9 not taken.

✓ Branch 10 taken 7436921 times.

✗ Branch 11 not taken.

✓ Branch 12 taken 370 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 1428 times.

✗ Branch 15 not taken.

✓ Branch 16 taken 141298 times.

✓ Branch 17 taken 150340 times.

✓ Branch 18 taken 136936 times.

✓ Branch 19 taken 155155 times.

✓ Branch 20 taken 5 times.

✓ Branch 21 taken 4 times.

✓ Branch 22 taken 20 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 100 times.

✗ Branch 25 not taken.

✓ Branch 26 taken 98 times.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✓ Branch 30 taken 30 times.

✓ Branch 31 taken 54 times.

✓ Branch 32 taken 30 times.

✓ Branch 33 taken 54 times.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

523830696

int mblen = mb_wc(&wc, sbeg, send);

1301

31/36

✓ Branch 0 taken 173831771 times.

✓ Branch 1 taken 173594221 times.

✓ Branch 2 taken 80672266 times.

✓ Branch 3 taken 80144321 times.

✓ Branch 4 taken 2136 times.

✓ Branch 5 taken 99890 times.

✓ Branch 6 taken 648 times.

✓ Branch 7 taken 26298 times.

✓ Branch 8 taken 113777 times.

✓ Branch 9 taken 7308007 times.

✓ Branch 10 taken 113591 times.

✓ Branch 11 taken 7323330 times.

✓ Branch 12 taken 232 times.

✓ Branch 13 taken 138 times.

✓ Branch 14 taken 418 times.

✓ Branch 15 taken 1010 times.

✓ Branch 16 taken 141298 times.

✓ Branch 17 taken 150340 times.

✓ Branch 18 taken 136936 times.

✓ Branch 19 taken 155155 times.

✓ Branch 20 taken 5 times.

✓ Branch 21 taken 4 times.

✓ Branch 22 taken 20 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 19 times.

✓ Branch 25 taken 81 times.

✓ Branch 26 taken 19 times.

✓ Branch 27 taken 79 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✓ Branch 30 taken 30 times.

✓ Branch 31 taken 54 times.

✓ Branch 32 taken 30 times.

✓ Branch 33 taken 54 times.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

523816178

if (mblen <= 0) {

1302

255013196

++weight_lv;

1303

19/36

✓ Branch 0 taken 173389403 times.

✓ Branch 1 taken 442368 times.

✓ Branch 2 taken 80229898 times.

✓ Branch 3 taken 442368 times.

✓ Branch 4 taken 2136 times.

✗ Branch 5 not taken.

✓ Branch 6 taken 648 times.

✗ Branch 7 not taken.

✓ Branch 8 taken 113778 times.

✗ Branch 9 not taken.

✓ Branch 10 taken 112565 times.

✓ Branch 11 taken 1026 times.

✓ Branch 12 taken 232 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 418 times.

✗ Branch 15 not taken.

✓ Branch 16 taken 141298 times.

✗ Branch 17 not taken.

✓ Branch 18 taken 136936 times.

✗ Branch 19 not taken.

✓ Branch 20 taken 5 times.

✗ Branch 21 not taken.

✓ Branch 22 taken 20 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 19 times.

✗ Branch 25 not taken.

✓ Branch 26 taken 19 times.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✓ Branch 30 taken 30 times.

✗ Branch 31 not taken.

✓ Branch 32 taken 30 times.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

503452146

if (sbeg >= send) return -1; /* No more bytes, end of line reached */

1304

/*

1305

There are some more bytes left. Non-positive mb_len means that

1306

we got an incomplete or a bad byte sequence. Consume mbminlen bytes.

1307

*/

1308

3/36

✗ Branch 0 not taken.

✓ Branch 1 taken 442368 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 442368 times.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✓ Branch 11 taken 1026 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

885761

if ((sbeg += cs->mbminlen) > send) {

1309

/* For safety purposes don't go beyond the string range. */

1310

✗

sbeg = send;

1311

}

1312

/*

1313

Treat every complete or incomplete mbminlen unit as a weight which is

1314

greater than weight for any possible normal character.

1315

0xFFFF is greater than any possible weight in the UCA weight table.

1316

*/

1317

885761

return 0xFFFF;

1318

}

1319

1320

268802982

sbeg += mblen;

1321

268802982

char_index++;

1322

18/36

✓ Branch 0 taken 144703492 times.

✓ Branch 1 taken 28890729 times.

✓ Branch 2 taken 72351751 times.

✓ Branch 3 taken 7792570 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 99890 times.

✗ Branch 6 not taken.

✓ Branch 7 taken 26298 times.

✗ Branch 8 not taken.

✓ Branch 9 taken 7308007 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 7323330 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 138 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 1010 times.

✗ Branch 16 not taken.

✓ Branch 17 taken 150340 times.

✓ Branch 18 taken 1 times.

✓ Branch 19 taken 155154 times.

✗ Branch 20 not taken.

✓ Branch 21 taken 4 times.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✓ Branch 25 taken 81 times.

✗ Branch 26 not taken.

✓ Branch 27 taken 79 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✓ Branch 31 taken 54 times.

✗ Branch 32 not taken.

✓ Branch 33 taken 54 times.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

268802982

if (wc > uca->maxchar) {

1323

/* Return 0xFFFD as weight for all characters outside BMP */

1324

217055244

wbeg = nochar;

1325

217055244

wbeg_stride = 0;

1326

217055244

return 0xFFFD;

1327

}

1328

1329

24/36

✓ Branch 0 taken 4798552 times.

✓ Branch 1 taken 24092177 times.

✓ Branch 2 taken 1161710 times.

✓ Branch 3 taken 6630860 times.

✓ Branch 4 taken 35488 times.

✓ Branch 5 taken 64402 times.

✓ Branch 6 taken 8913 times.

✓ Branch 7 taken 17385 times.

✓ Branch 8 taken 4441 times.

✓ Branch 9 taken 7309177 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 7308254 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 138 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 1010 times.

✓ Branch 16 taken 4263 times.

✓ Branch 17 taken 146077 times.

✓ Branch 18 taken 4468 times.

✓ Branch 19 taken 150686 times.

✗ Branch 20 not taken.

✓ Branch 21 taken 4 times.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✓ Branch 24 taken 2 times.

✓ Branch 25 taken 79 times.

✓ Branch 26 taken 2 times.

✓ Branch 27 taken 77 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✓ Branch 31 taken 54 times.

✗ Branch 32 not taken.

✓ Branch 33 taken 54 times.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

51747738

if (my_uca_have_contractions(uca)) {

1330

const uint16 *cweight;

1331

/*

1332

If we have scanned a code point which can have previous context,

1333

and there were some more code point already before,

1334

then verify that {prev_char, wc} together form

1335

a real previous context pair.

1336

Note, we support only 2-character long sequences with previous

1337

context at the moment. CLDR does not have longer sequences.

1338

*/

1339

6004265

if (my_uca_can_be_previous_context_tail(uca->contraction_flags, wc) &&

1340

4/72

✓ Branch 0 taken 56 times.

✓ Branch 1 taken 8 times.

✓ Branch 2 taken 40 times.

✓ Branch 3 taken 16 times.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

120

wbeg != nochar && /* if not the very first character */

1341

56

my_uca_can_be_previous_context_head(uca->contraction_flags,

1342

22/72

✓ Branch 0 taken 64 times.

✓ Branch 1 taken 4798488 times.

✓ Branch 2 taken 40 times.

✓ Branch 3 taken 4798512 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 1161710 times.

✗ Branch 6 not taken.

✓ Branch 7 taken 1161710 times.

✗ Branch 8 not taken.

✓ Branch 9 taken 35488 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 35488 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 8913 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 8913 times.

✗ Branch 16 not taken.

✓ Branch 17 taken 14030 times.

✗ Branch 18 not taken.

✓ Branch 19 taken 14030 times.

✗ Branch 20 not taken.

✓ Branch 21 taken 14836 times.

✗ Branch 22 not taken.

✓ Branch 23 taken 14836 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✓ Branch 33 taken 4263 times.

✗ Branch 34 not taken.

✓ Branch 35 taken 4263 times.

✗ Branch 36 not taken.

✓ Branch 37 taken 4468 times.

✗ Branch 38 not taken.

✓ Branch 39 taken 4468 times.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✓ Branch 49 taken 2 times.

✗ Branch 50 not taken.

✓ Branch 51 taken 2 times.

✗ Branch 52 not taken.

✓ Branch 53 taken 2 times.

✗ Branch 54 not taken.

✓ Branch 55 taken 2 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

6042328

prev_char) &&

1343

2/72

✓ Branch 0 taken 40 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 40 times.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

80

(cweight = previous_context_find(prev_char, wc))) {

1344

40

prev_char = 0; /* Clear for the next character */

1345

40

return *cweight;

1346

18/36

✓ Branch 0 taken 5210 times.

✓ Branch 1 taken 4793302 times.

✓ Branch 2 taken 1050 times.

✓ Branch 3 taken 1160660 times.

✓ Branch 4 taken 1280 times.

✓ Branch 5 taken 34208 times.

✓ Branch 6 taken 320 times.

✓ Branch 7 taken 8593 times.

✓ Branch 8 taken 1425 times.

✓ Branch 9 taken 12605 times.

✓ Branch 10 taken 1159 times.

✓ Branch 11 taken 13677 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 540 times.

✓ Branch 17 taken 3723 times.

✓ Branch 18 taken 421 times.

✓ Branch 19 taken 4047 times.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✓ Branch 25 taken 2 times.

✗ Branch 26 not taken.

✓ Branch 27 taken 2 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

6042224

} else if (my_uca_can_be_contraction_head(uca->contraction_flags, wc)) {

1347

/* Check if wc starts a contraction */

1348

size_t chars_skipped;

1349

24/72

✓ Branch 0 taken 5210 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 1007 times.

✓ Branch 3 taken 4203 times.

✓ Branch 4 taken 1050 times.

✗ Branch 5 not taken.

✓ Branch 6 taken 36 times.

✓ Branch 7 taken 1014 times.

✓ Branch 8 taken 1280 times.

✗ Branch 9 not taken.

✓ Branch 10 taken 244 times.

✓ Branch 11 taken 1036 times.

✓ Branch 12 taken 320 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 61 times.

✓ Branch 15 taken 259 times.

✓ Branch 16 taken 1425 times.

✗ Branch 17 not taken.

✓ Branch 18 taken 508 times.

✓ Branch 19 taken 917 times.

✓ Branch 20 taken 1159 times.

✗ Branch 21 not taken.

✓ Branch 22 taken 583 times.

✓ Branch 23 taken 576 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✓ Branch 32 taken 540 times.

✗ Branch 33 not taken.

✓ Branch 34 taken 59 times.

✓ Branch 35 taken 481 times.

✓ Branch 36 taken 421 times.

✗ Branch 37 not taken.

✓ Branch 38 taken 121 times.

✓ Branch 39 taken 300 times.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

11405

if ((cweight = contraction_find(wc, &chars_skipped))) {

1350

2619

char_index += chars_skipped;

1351

2619

return *cweight;

1352

}

1353

}

1354

6039605

prev_char = wc;

1355

}

1356

1357

/* Process single code point */

1358

51760039

uint page = wc >> 8;

1359

51760039

uint code = wc & 0xFF;

1360

1361

/* If weight page for wc does not exist, then calculate algorithmically */

1362

51760039

const uint16 *wpage = uca->weights[page];

1363

29/72

✓ Branch 0 taken 23725887 times.

✓ Branch 1 taken 5163795 times.

✓ Branch 2 taken 23301951 times.

✓ Branch 3 taken 423936 times.

✓ Branch 4 taken 6589443 times.

✓ Branch 5 taken 1203091 times.

✓ Branch 6 taken 6483459 times.

✓ Branch 7 taken 105984 times.

✗ Branch 8 not taken.

✓ Branch 9 taken 99646 times.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✓ Branch 13 taken 26237 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 372 times.

✓ Branch 17 taken 7322327 times.

✓ Branch 18 taken 372 times.

✗ Branch 19 not taken.

✓ Branch 20 taken 357 times.

✓ Branch 21 taken 7322150 times.

✓ Branch 22 taken 357 times.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✓ Branch 25 taken 138 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✓ Branch 29 taken 1010 times.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✓ Branch 32 taken 89615 times.

✓ Branch 33 taken 60666 times.

✓ Branch 34 taken 89615 times.

✗ Branch 35 not taken.

✓ Branch 36 taken 89612 times.

✓ Branch 37 taken 65421 times.

✓ Branch 38 taken 89612 times.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✓ Branch 41 taken 4 times.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✓ Branch 49 taken 81 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✓ Branch 53 taken 79 times.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✓ Branch 61 taken 54 times.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✓ Branch 65 taken 54 times.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

82255325

if (!wpage) return next_implicit(wc);

1364

1365

/* Calculate pointer to wc's weight, using page and offset */

1366

21264753

wbeg = wpage + code * uca->lengths[page];

1367

21264753

wbeg_stride = UCA900_DISTANCE_BETWEEN_WEIGHTS;

1368

23/36

✓ Branch 0 taken 170235 times.

✓ Branch 1 taken 4993560 times.

✓ Branch 2 taken 35511 times.

✓ Branch 3 taken 1167580 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 99646 times.

✗ Branch 6 not taken.

✓ Branch 7 taken 26237 times.

✓ Branch 8 taken 362 times.

✓ Branch 9 taken 7321965 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 7323166 times.

✓ Branch 12 taken 2 times.

✓ Branch 13 taken 136 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 1010 times.

✓ Branch 16 taken 1330 times.

✓ Branch 17 taken 59336 times.

✓ Branch 18 taken 1365 times.

✓ Branch 19 taken 64056 times.

✓ Branch 20 taken 1 times.

✓ Branch 21 taken 3 times.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✓ Branch 24 taken 2 times.

✓ Branch 25 taken 79 times.

✗ Branch 26 not taken.

✓ Branch 27 taken 79 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✓ Branch 31 taken 54 times.

✗ Branch 32 not taken.

✓ Branch 33 taken 54 times.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

21264753

} while (!wbeg[0]); /* Skip ignorable code points */

1369

1370

21056961

return *wbeg++;

1371

}

1372

1373

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1374

79843923616

inline int uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::more_weight() {

1375

/*

1376

Check if the weights for the previous code point have been

1377

already fully scanned. If no, return the first non-zero

weight.

*/

4/4

✓ Branch 0 taken 418595692 times.

✓ Branch 1 taken 39712863105 times.

✓ Branch 2 taken 209496989 times.

✓ Branch 3 taken 209098703 times.

80262917594

while (num_of_ce_left != 0 && *wbeg == 0) {

1382

418993978

wbeg += wbeg_stride;

1383

418993978

--num_of_ce_left;

1384

}

1385

2/2

✓ Branch 0 taken 209098703 times.

✓ Branch 1 taken 39712863105 times.

79843923616

if (num_of_ce_left != 0) {

1386

418197406

uint16 rtn = *wbeg;

1387

418197406

wbeg += wbeg_stride;

1388

418197406

--num_of_ce_left;

1389

418197406

return rtn; /* return the next weight from expansion */

1390

}

1391

79425726210

return -1;

1392

}

1393

1394

3341369

static inline bool is_hiragana_char(my_wc_t wc) {

1395

4/4

✓ Branch 0 taken 3301899 times.

✓ Branch 1 taken 39470 times.

✓ Branch 2 taken 2449 times.

✓ Branch 3 taken 3299450 times.

3341369

return wc >= 0x3041 && wc <= 0x3096;

1396

}

1397

1398

3343491

static inline bool is_katakana_char(my_wc_t wc) {

1399

6/6

✓ Branch 0 taken 3301534 times.

✓ Branch 1 taken 41957 times.

✓ Branch 2 taken 3299777 times.

✓ Branch 3 taken 1757 times.

✓ Branch 4 taken 3146358 times.

✓ Branch 5 taken 195376 times.

6489849

return (wc >= 0x30A1 && wc <= 0x30FA) || // Full width katakana

1400

2/2

✓ Branch 0 taken 336 times.

✓ Branch 1 taken 3146022 times.

6489849

(wc >= 0xFF66 && wc <= 0xFF9D); // Half width katakana

1401

}

1402

1403

3341370

static inline bool is_katakana_iteration(my_wc_t wc) {

1404

4/4

✓ Branch 0 taken 3341362 times.

✓ Branch 1 taken 8 times.

✓ Branch 2 taken 6 times.

✓ Branch 3 taken 3341356 times.

3341370

return wc == 0x30FD || wc == 0x30FE;

1405

}

1406

1407

3338920

static inline bool is_hiragana_iteration(my_wc_t wc) {

1408

4/4

✓ Branch 0 taken 3338912 times.

✓ Branch 1 taken 8 times.

✓ Branch 2 taken 6 times.

✓ Branch 3 taken 3338906 times.

3338920

return wc == 0x309D || wc == 0x309E;

1409

}

1410

1411

3340713

static inline bool is_ja_length_mark(my_wc_t wc) { return wc == 0x30FC; }

1412

1413

/**

1414

Return quaternary weight when running for that level.

1415

1416

@retval 0 - Do not return quaternary weight.

1417

@retval others - Quaternary weight for this character.

1418

*/

1419

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1420

ALWAYS_INLINE int

1421

uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::handle_ja_contraction_quat_wt() {

1422

/*

1423

For Japanese, only weight shift rule and previous context rule is

1424

defined. And in previous context rules, the first character is always

1425

katakana / hiragana, and the second character is always iteration or

1426

length mark. The quaternary weight of iteration / length mark is

1427

same as the first character. So has_quaternary_weight is always true.

1428

For how we return quaternary weight, please refer to the comment in

1429

handle_ja_common_quat_wt().

1430

*/

1431

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 224 times.

✓ Branch 5 taken 1389 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 14 times.

✓ Branch 13 taken 42 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 12 times.

✓ Branch 23 taken 45 times.

✓ Branch 24 taken 12 times.

✓ Branch 25 taken 45 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

1783

if (weight_lv == 3) {

1432

262

wbeg = nochar;

1433

262

num_of_ce_left = 0;

1434

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 214 times.

✓ Branch 5 taken 10 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 8 times.

✓ Branch 13 taken 6 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 6 times.

✓ Branch 23 taken 6 times.

✓ Branch 24 taken 6 times.

✓ Branch 25 taken 6 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

262

if (is_katakana_char(prev_char)) {

1435

234

return JA_KATA_QUAT_WEIGHT;

1436

4/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 10 times.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 6 times.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 6 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 6 times.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

28

} else if (is_hiragana_char(prev_char)) {

1437

28

return JA_HIRA_QUAT_WEIGHT;

1438

}

1439

}

1440

1521

return 0;

}

/**

Check whether quaternary weight is needed for character with Japanese

1445

kana-sensitive collation. If it is, return quaternary weight when running

1446

for that level.

1447

1448

@retval 0 - Quaternary weight check is done.

1449

@retval -1 - There is no quaternary weight for this character.

1450

@retval others - Quaternary weight for this character.

1451

*/

1452

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1453

ALWAYS_INLINE int

1454

uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::handle_ja_common_quat_wt(

1455

my_wc_t wc) {

1456

/*

1457

For Japanese kana-sensitive collation, we detect whether quaternary

1458

weight is necessary when scanning for the first level of weight.

1459

If it is, the quaternary weight will be returned for katakana /

1460

hiragana later.

1461

*/

1462

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 3337783 times.

✓ Branch 5 taken 7338 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 740 times.

✓ Branch 13 taken 39 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 80 times.

✓ Branch 23 taken 54 times.

✓ Branch 24 taken 79 times.

✓ Branch 25 taken 54 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

3346167

if (weight_lv == 0 && !has_quaternary_weight) {

1463

14/56

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✓ Branch 8 taken 3337288 times.

✓ Branch 9 taken 6 times.

✓ Branch 10 taken 3336717 times.

✓ Branch 11 taken 571 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✓ Branch 24 taken 724 times.

✗ Branch 25 not taken.

✓ Branch 26 taken 712 times.

✓ Branch 27 taken 12 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✓ Branch 44 taken 59 times.

✗ Branch 45 not taken.

✓ Branch 46 taken 31 times.

✓ Branch 47 taken 28 times.

✓ Branch 48 taken 56 times.

✓ Branch 49 taken 1 times.

✓ Branch 50 taken 31 times.

✓ Branch 51 taken 25 times.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

10014943

if (is_katakana_char(wc) || is_katakana_iteration(wc) ||

1464

30/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 3337294 times.

✓ Branch 17 taken 489 times.

✓ Branch 18 taken 3336711 times.

✓ Branch 19 taken 6 times.

✓ Branch 20 taken 5 times.

✓ Branch 21 taken 3336706 times.

✓ Branch 22 taken 1077 times.

✓ Branch 23 taken 3336706 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 724 times.

✓ Branch 49 taken 16 times.

✓ Branch 50 taken 712 times.

✗ Branch 51 not taken.

✓ Branch 52 taken 4 times.

✓ Branch 53 taken 708 times.

✓ Branch 54 taken 32 times.

✓ Branch 55 taken 708 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 59 times.

✓ Branch 89 taken 21 times.

✓ Branch 90 taken 30 times.

✓ Branch 91 taken 1 times.

✓ Branch 92 taken 3 times.

✓ Branch 93 taken 27 times.

✓ Branch 94 taken 53 times.

✓ Branch 95 taken 27 times.

✓ Branch 96 taken 57 times.

✓ Branch 97 taken 22 times.

✓ Branch 98 taken 31 times.

✗ Branch 99 not taken.

✓ Branch 100 taken 8 times.

✓ Branch 101 taken 23 times.

✓ Branch 102 taken 56 times.

✓ Branch 103 taken 23 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

13352427

is_hiragana_char(wc) || is_hiragana_iteration(wc) ||

1465

3337484

is_ja_length_mark(wc))

1466

1218

has_quaternary_weight = true;

1467

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 4284 times.

✓ Branch 5 taken 6691499 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 83 times.

✓ Branch 13 taken 1597 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 90 times.

✓ Branch 23 taken 279 times.

✓ Branch 24 taken 90 times.

✓ Branch 25 taken 279 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

6698201

} else if (weight_lv == 3) {

1468

4547

wbeg = nochar;

1469

4547

num_of_ce_left = 0;

1470

30/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 3081 times.

✓ Branch 17 taken 1203 times.

✓ Branch 18 taken 3075 times.

✓ Branch 19 taken 6 times.

✓ Branch 20 taken 5 times.

✓ Branch 21 taken 3070 times.

✓ Branch 22 taken 1214 times.

✓ Branch 23 taken 3070 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 43 times.

✓ Branch 49 taken 40 times.

✓ Branch 50 taken 43 times.

✗ Branch 51 not taken.

✓ Branch 52 taken 4 times.

✓ Branch 53 taken 39 times.

✓ Branch 54 taken 44 times.

✓ Branch 55 taken 39 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 57 times.

✓ Branch 89 taken 33 times.

✓ Branch 90 taken 57 times.

✗ Branch 91 not taken.

✓ Branch 92 taken 3 times.

✓ Branch 93 taken 54 times.

✓ Branch 94 taken 36 times.

✓ Branch 95 taken 54 times.

✓ Branch 96 taken 55 times.

✓ Branch 97 taken 35 times.

✓ Branch 98 taken 54 times.

✓ Branch 99 taken 1 times.

✓ Branch 100 taken 3 times.

✓ Branch 101 taken 51 times.

✓ Branch 102 taken 39 times.

✓ Branch 103 taken 51 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

7776

if (is_katakana_char(wc) || is_katakana_iteration(wc) ||

1471

3229

is_ja_length_mark(wc)) {

1472

1333

return JA_KATA_QUAT_WEIGHT;

1473

22/84

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 1398 times.

✓ Branch 13 taken 1672 times.

✓ Branch 14 taken 6 times.

✓ Branch 15 taken 1392 times.

✓ Branch 16 taken 1678 times.

✓ Branch 17 taken 1392 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✓ Branch 36 taken 12 times.

✓ Branch 37 taken 27 times.

✗ Branch 38 not taken.

✓ Branch 39 taken 12 times.

✓ Branch 40 taken 27 times.

✓ Branch 41 taken 12 times.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✓ Branch 66 taken 10 times.

✓ Branch 67 taken 44 times.

✓ Branch 68 taken 1 times.

✓ Branch 69 taken 9 times.

✓ Branch 70 taken 45 times.

✓ Branch 71 taken 9 times.

✓ Branch 72 taken 9 times.

✓ Branch 73 taken 42 times.

✗ Branch 74 not taken.

✓ Branch 75 taken 9 times.

✓ Branch 76 taken 42 times.

✓ Branch 77 taken 9 times.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

3214

} else if (is_hiragana_char(wc) || is_hiragana_iteration(wc)) {

1474

1792

return JA_HIRA_QUAT_WEIGHT;

1475

}

1476

1422

return -1;

1477

}

1478

10032336

return 0;

1479

}

1480

1481

// Generic version that can handle any number of levels.

1482

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1483

ALWAYS_INLINE int uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::next_raw() {

1484

20171736869

int remain_weight = more_weight();

1485

36/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 3577324 times.

✓ Branch 17 taken 20053073 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 103283319 times.

✓ Branch 21 taken 600846594 times.

✗ Branch 22 not taken.

✓ Branch 23 taken 20040882 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 13362597 times.

✓ Branch 28 taken 92111265 times.

✓ Branch 29 taken 186919638 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 94401968 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 56 times.

✓ Branch 49 taken 2610 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 108 times.

✓ Branch 53 taken 61983 times.

✗ Branch 54 not taken.

✓ Branch 55 taken 459 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✓ Branch 59 taken 56 times.

✓ Branch 60 taken 45 times.

✓ Branch 61 taken 19281 times.

✗ Branch 62 not taken.

✓ Branch 63 taken 375695071 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 67 times.

✓ Branch 89 taken 659 times.

✓ Branch 90 taken 67 times.

✓ Branch 91 taken 659 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 297 times.

✓ Branch 95 taken 8755 times.

✓ Branch 96 taken 279 times.

✓ Branch 97 taken 8773 times.

✗ Branch 98 not taken.

✓ Branch 99 taken 96 times.

✓ Branch 100 taken 93 times.

✓ Branch 101 taken 16113 times.

✓ Branch 102 taken 93 times.

✓ Branch 103 taken 16113 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 8214 times.

✓ Branch 107 taken 19130533695 times.

✓ Branch 108 taken 8258 times.

✓ Branch 109 taken 19177966079 times.

✗ Branch 110 not taken.

✓ Branch 111 taken 4116 times.

39818948755

if (remain_weight >= 0) return remain_weight;

1486

1487

do {

1488

39712045605

my_wc_t wc = 0;

1489

1490

/* Get next code point */

1491

44/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 10009965 times.

✓ Branch 17 taken 10053611 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 300260109 times.

✓ Branch 21 taken 300859818 times.

✓ Branch 22 taken 10013990 times.

✓ Branch 23 taken 10036096 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 6677229 times.

✓ Branch 27 taken 6692036 times.

✓ Branch 28 taken 93427849 times.

✓ Branch 29 taken 93606617 times.

✓ Branch 30 taken 39165543 times.

✓ Branch 31 taken 55241660 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 146 times.

✓ Branch 49 taken 2476 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 270 times.

✓ Branch 53 taken 61713 times.

✓ Branch 54 taken 153 times.

✓ Branch 55 taken 306 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✓ Branch 58 taken 28 times.

✓ Branch 59 taken 28 times.

✓ Branch 60 taken 84 times.

✓ Branch 61 taken 19197 times.

✓ Branch 62 taken 151912324 times.

✓ Branch 63 taken 223783187 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 162 times.

✓ Branch 89 taken 506 times.

✓ Branch 90 taken 163 times.

✓ Branch 91 taken 505 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 2119 times.

✓ Branch 95 taken 6642 times.

✓ Branch 96 taken 2172 times.

✓ Branch 97 taken 6601 times.

✓ Branch 98 taken 39 times.

✓ Branch 99 taken 57 times.

✓ Branch 100 taken 1816 times.

✓ Branch 101 taken 14298 times.

✓ Branch 102 taken 1825 times.

✓ Branch 103 taken 14288 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 368824991 times.

✓ Branch 107 taken 18795301959 times.

✓ Branch 108 taken 367868608 times.

✓ Branch 109 taken 18868170303 times.

✓ Branch 110 taken 1599 times.

✓ Branch 111 taken 2517 times.

39712045605

int mblen = mb_wc(&wc, sbeg, send);

1492

44/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 10028397 times.

✓ Branch 17 taken 10035179 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 300813069 times.

✓ Branch 21 taken 300306858 times.

✓ Branch 22 taken 10032422 times.

✓ Branch 23 taken 10017664 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 6689517 times.

✓ Branch 27 taken 6679748 times.

✓ Branch 28 taken 93599881 times.

✓ Branch 29 taken 93434585 times.

✓ Branch 30 taken 39171687 times.

✓ Branch 31 taken 55235516 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 146 times.

✓ Branch 49 taken 2476 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 270 times.

✓ Branch 53 taken 61713 times.

✓ Branch 54 taken 153 times.

✓ Branch 55 taken 306 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✓ Branch 58 taken 28 times.

✓ Branch 59 taken 28 times.

✓ Branch 60 taken 84 times.

✓ Branch 61 taken 19197 times.

✓ Branch 62 taken 151912310 times.

✓ Branch 63 taken 223783201 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 162 times.

✓ Branch 89 taken 506 times.

✓ Branch 90 taken 163 times.

✓ Branch 91 taken 505 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 2119 times.

✓ Branch 95 taken 6642 times.

✓ Branch 96 taken 2172 times.

✓ Branch 97 taken 6601 times.

✓ Branch 98 taken 39 times.

✓ Branch 99 taken 57 times.

✓ Branch 100 taken 1816 times.

✓ Branch 101 taken 14298 times.

✓ Branch 102 taken 1825 times.

✓ Branch 103 taken 14288 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 368811613 times.

✓ Branch 107 taken 18795315337 times.

✓ Branch 108 taken 367858504 times.

✓ Branch 109 taken 18868180407 times.

✓ Branch 110 taken 1599 times.

✓ Branch 111 taken 2517 times.

39712045605

if (mblen <= 0) {

1493

if (LEVELS_FOR_COMPARE == 1) {

1494

1021355678

++weight_lv;

1495

1757075983

return -1;

1496

}

1497

1498

30/84

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 10027641 times.

✓ Branch 13 taken 756 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 200542232 times.

✓ Branch 17 taken 100270837 times.

✓ Branch 18 taken 6688324 times.

✓ Branch 19 taken 3344098 times.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 3344797 times.

✓ Branch 23 taken 3344720 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✓ Branch 36 taken 114 times.

✓ Branch 37 taken 32 times.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✓ Branch 40 taken 180 times.

✓ Branch 41 taken 90 times.

✓ Branch 42 taken 102 times.

✓ Branch 43 taken 51 times.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✓ Branch 46 taken 14 times.

✓ Branch 47 taken 14 times.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✓ Branch 66 taken 131 times.

✓ Branch 67 taken 31 times.

✓ Branch 68 taken 132 times.

✓ Branch 69 taken 31 times.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✓ Branch 72 taken 1866 times.

✓ Branch 73 taken 253 times.

✓ Branch 74 taken 1914 times.

✓ Branch 75 taken 258 times.

✓ Branch 76 taken 30 times.

✓ Branch 77 taken 9 times.

✓ Branch 78 taken 931 times.

✓ Branch 79 taken 885 times.

✓ Branch 80 taken 940 times.

✓ Branch 81 taken 885 times.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

327572298

if (++weight_lv < LEVELS_FOR_COMPARE) {

1499

4/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 10027641 times.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 114 times.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 131 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 132 times.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

10028018

if (LEVELS_FOR_COMPARE == 4 && cs->coll_param == &ja_coll_param) {

1500

// Return directly if we don't have quaternary weight.

1501

16/56

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✓ Branch 8 taken 3342459 times.

✓ Branch 9 taken 6685182 times.

✓ Branch 10 taken 3341646 times.

✓ Branch 11 taken 813 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✓ Branch 24 taken 38 times.

✓ Branch 25 taken 76 times.

✓ Branch 26 taken 6 times.

✓ Branch 27 taken 32 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✓ Branch 44 taken 43 times.

✓ Branch 45 taken 88 times.

✓ Branch 46 taken 3 times.

✓ Branch 47 taken 40 times.

✓ Branch 48 taken 43 times.

✓ Branch 49 taken 89 times.

✓ Branch 50 taken 3 times.

✓ Branch 51 taken 40 times.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

10028018

if (weight_lv == 3 && !has_quaternary_weight) return -1;

1502

}

1503

/*

1504

Restart scanning from the beginning of the string, and add

1505

a level separator.

1506

*/

1507

217267690

sbeg = sbeg_dup;

1508

217267690

return 0;

1509

}

1510

1511

// If we don't have any more levels left, we're done.

1512

106962950

return -1;

1513

}

1514

1515

38363117629

sbeg += mblen;

1516

22/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✓ Branch 17 taken 10035179 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✓ Branch 21 taken 300306858 times.

✗ Branch 22 not taken.

✓ Branch 23 taken 10017664 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 6679748 times.

✗ Branch 28 not taken.

✓ Branch 29 taken 93434585 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 55235516 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✓ Branch 49 taken 2476 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✓ Branch 53 taken 61713 times.

✗ Branch 54 not taken.

✓ Branch 55 taken 306 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✓ Branch 59 taken 28 times.

✗ Branch 60 not taken.

✓ Branch 61 taken 19197 times.

✗ Branch 62 not taken.

✓ Branch 63 taken 223783201 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✓ Branch 89 taken 506 times.

✗ Branch 90 not taken.

✓ Branch 91 taken 505 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✓ Branch 95 taken 6642 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 6601 times.

✗ Branch 98 not taken.

✓ Branch 99 taken 57 times.

✗ Branch 100 not taken.

✓ Branch 101 taken 14298 times.

✗ Branch 102 not taken.

✓ Branch 103 taken 14288 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✓ Branch 107 taken 18795315337 times.

✗ Branch 108 not taken.

✓ Branch 109 taken 18868180407 times.

✗ Branch 110 not taken.

✓ Branch 111 taken 2517 times.

38363117629

assert(wc <= uca->maxchar); // mb_wc() has already checked this.

1517

1518

31/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 10035179 times.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 130159215 times.

✓ Branch 21 taken 170147643 times.

✗ Branch 22 not taken.

✓ Branch 23 taken 10017664 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 6679748 times.

✓ Branch 28 taken 36705833 times.

✓ Branch 29 taken 56728752 times.

✓ Branch 30 taken 604 times.

✓ Branch 31 taken 55235078 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 2476 times.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 26676 times.

✓ Branch 53 taken 35037 times.

✗ Branch 54 not taken.

✓ Branch 55 taken 306 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✓ Branch 59 taken 28 times.

✓ Branch 60 taken 7518 times.

✓ Branch 61 taken 11679 times.

✓ Branch 62 taken 18 times.

✓ Branch 63 taken 223782848 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 506 times.

✗ Branch 89 not taken.

✓ Branch 90 taken 505 times.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 121 times.

✓ Branch 95 taken 6521 times.

✓ Branch 96 taken 121 times.

✓ Branch 97 taken 6480 times.

✗ Branch 98 not taken.

✓ Branch 99 taken 57 times.

✗ Branch 100 not taken.

✓ Branch 101 taken 14298 times.

✗ Branch 102 not taken.

✓ Branch 103 taken 14288 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 51445150 times.

✓ Branch 107 taken 18750560320 times.

✗ Branch 108 not taken.

✓ Branch 109 taken 18815643420 times.

✗ Branch 110 not taken.

✓ Branch 111 taken 2517 times.

38363117629

if (my_uca_have_contractions(uca)) {

1519

const uint16 *cweight;

1520

/*

1521

If we have scanned a code point which can have previous context,

1522

and there were some more code points already before,

1523

then verify that {prev_char, wc} together form

1524

a real previous context pair.

1525

Note, we support only 2-character long sequences with previous

1526

context at the moment. CLDR does not have longer sequences.

1527

CLDR doesn't have previous context rule whose first character is

1528

0x0000, so the initial value (0) of prev_char won't break the logic.

1529

*/

1530

12/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 1613 times.

✓ Branch 17 taken 12263 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 639 times.

✓ Branch 21 taken 12240 times.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 56 times.

✓ Branch 49 taken 16 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 57 times.

✓ Branch 89 taken 16 times.

✓ Branch 90 taken 57 times.

✓ Branch 91 taken 21 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✓ Branch 95 taken 3 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 3 times.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

213602186

if (my_uca_can_be_previous_context_tail(uca->contraction_flags, wc) &&

1531

26984

my_uca_can_be_previous_context_head(uca->contraction_flags,

1532

36/224

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✓ Branch 32 taken 13876 times.

✓ Branch 33 taken 10021303 times.

✓ Branch 34 taken 1613 times.

✓ Branch 35 taken 10033566 times.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✓ Branch 40 taken 12879 times.

✓ Branch 41 taken 130146336 times.

✓ Branch 42 taken 639 times.

✓ Branch 43 taken 130158576 times.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✓ Branch 57 taken 36705833 times.

✗ Branch 58 not taken.

✓ Branch 59 taken 36705833 times.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✓ Branch 96 taken 72 times.

✓ Branch 97 taken 2404 times.

✓ Branch 98 taken 56 times.

✓ Branch 99 taken 2420 times.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✓ Branch 105 taken 26676 times.

✗ Branch 106 not taken.

✓ Branch 107 taken 26676 times.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✓ Branch 121 taken 7518 times.

✗ Branch 122 not taken.

✓ Branch 123 taken 7518 times.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✗ Branch 148 not taken.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✓ Branch 176 taken 73 times.

✓ Branch 177 taken 433 times.

✓ Branch 178 taken 57 times.

✓ Branch 179 taken 449 times.

✓ Branch 180 taken 78 times.

✓ Branch 181 taken 427 times.

✓ Branch 182 taken 57 times.

✓ Branch 183 taken 448 times.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✓ Branch 188 taken 3 times.

✓ Branch 189 taken 118 times.

✗ Branch 190 not taken.

✓ Branch 191 taken 121 times.

✓ Branch 192 taken 3 times.

✓ Branch 193 taken 118 times.

✗ Branch 194 not taken.

✓ Branch 195 taken 121 times.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✓ Branch 213 taken 5140 times.

✗ Branch 214 not taken.

✓ Branch 215 taken 5140 times.

✗ Branch 216 not taken.

✓ Branch 217 taken 5231 times.

✗ Branch 218 not taken.

✓ Branch 219 taken 5231 times.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

176975505

prev_char) &&

1533

10/224

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✓ Branch 32 taken 1613 times.

✗ Branch 33 not taken.

✓ Branch 34 taken 1613 times.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✓ Branch 40 taken 639 times.

✗ Branch 41 not taken.

✓ Branch 42 taken 639 times.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✓ Branch 96 taken 56 times.

✗ Branch 97 not taken.

✓ Branch 98 taken 56 times.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✗ Branch 148 not taken.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✓ Branch 176 taken 57 times.

✗ Branch 177 not taken.

✓ Branch 178 taken 57 times.

✗ Branch 179 not taken.

✓ Branch 180 taken 57 times.

✗ Branch 181 not taken.

✓ Branch 182 taken 57 times.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✗ Branch 212 not taken.

✗ Branch 213 not taken.

✗ Branch 214 not taken.

✗ Branch 215 not taken.

✗ Branch 216 not taken.

✗ Branch 217 not taken.

✗ Branch 218 not taken.

✗ Branch 219 not taken.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

4844

(cweight = previous_context_find(prev_char, wc))) {

1534

// For Japanese kana-sensitive collation.

1535

4/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 1613 times.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 56 times.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 57 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 57 times.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

1783

if (LEVELS_FOR_COMPARE == 4 && cs->coll_param == &ja_coll_param) {

1536

1783

int quat_wt = handle_ja_contraction_quat_wt();

1537

1783

prev_char = 0;

1538

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 224 times.

✓ Branch 5 taken 1389 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 14 times.

✓ Branch 13 taken 42 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 12 times.

✓ Branch 23 taken 45 times.

✓ Branch 24 taken 12 times.

✓ Branch 25 taken 45 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

1783

if (quat_wt > 0) return quat_wt;

1539

}

1540

2160

prev_char = 0; /* Clear for the next code point */

1541

2160

return *cweight;

1542

18/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✓ Branch 17 taken 10033566 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 145311 times.

✓ Branch 21 taken 130013265 times.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✓ Branch 28 taken 40577 times.

✓ Branch 29 taken 36665256 times.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✓ Branch 49 taken 2420 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 2421 times.

✓ Branch 53 taken 24255 times.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✓ Branch 60 taken 717 times.

✓ Branch 61 taken 6801 times.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✓ Branch 89 taken 449 times.

✗ Branch 90 not taken.

✓ Branch 91 taken 448 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✓ Branch 95 taken 121 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 121 times.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 354 times.

✓ Branch 107 taken 4786 times.

✓ Branch 108 taken 312 times.

✓ Branch 109 taken 4919 times.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

176946099

} else if (my_uca_can_be_contraction_head(uca->contraction_flags, wc)) {

1543

/* Check if wc starts a contraction */

1544

size_t chars_skipped; // Ignored.

1545

18/224

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✓ Branch 40 taken 145311 times.

✗ Branch 41 not taken.

✓ Branch 42 taken 402 times.

✓ Branch 43 taken 144909 times.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✓ Branch 56 taken 40577 times.

✗ Branch 57 not taken.

✓ Branch 58 taken 139 times.

✓ Branch 59 taken 40438 times.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✓ Branch 104 taken 2421 times.

✗ Branch 105 not taken.

✓ Branch 106 taken 108 times.

✓ Branch 107 taken 2313 times.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✗ Branch 112 not taken.

✗ Branch 113 not taken.

✗ Branch 114 not taken.

✗ Branch 115 not taken.

✗ Branch 116 not taken.

✗ Branch 117 not taken.

✗ Branch 118 not taken.

✗ Branch 119 not taken.

✓ Branch 120 taken 717 times.

✗ Branch 121 not taken.

✓ Branch 122 taken 36 times.

✓ Branch 123 taken 681 times.

✗ Branch 124 not taken.

✗ Branch 125 not taken.

✗ Branch 126 not taken.

✗ Branch 127 not taken.

✗ Branch 128 not taken.

✗ Branch 129 not taken.

✗ Branch 130 not taken.

✗ Branch 131 not taken.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✗ Branch 148 not taken.

✗ Branch 149 not taken.

✗ Branch 150 not taken.

✗ Branch 151 not taken.

✗ Branch 152 not taken.

✗ Branch 153 not taken.

✗ Branch 154 not taken.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

✗ Branch 160 not taken.

✗ Branch 161 not taken.

✗ Branch 162 not taken.

✗ Branch 163 not taken.

✗ Branch 164 not taken.

✗ Branch 165 not taken.

✗ Branch 166 not taken.

✗ Branch 167 not taken.

✗ Branch 168 not taken.

✗ Branch 169 not taken.

✗ Branch 170 not taken.

✗ Branch 171 not taken.

✗ Branch 172 not taken.

✗ Branch 173 not taken.

✗ Branch 174 not taken.

✗ Branch 175 not taken.

✗ Branch 176 not taken.

✗ Branch 177 not taken.

✗ Branch 178 not taken.

✗ Branch 179 not taken.

✗ Branch 180 not taken.

✗ Branch 181 not taken.

✗ Branch 182 not taken.

✗ Branch 183 not taken.

✗ Branch 184 not taken.

✗ Branch 185 not taken.

✗ Branch 186 not taken.

✗ Branch 187 not taken.

✗ Branch 188 not taken.

✗ Branch 189 not taken.

✗ Branch 190 not taken.

✗ Branch 191 not taken.

✗ Branch 192 not taken.

✗ Branch 193 not taken.

✗ Branch 194 not taken.

✗ Branch 195 not taken.

✗ Branch 196 not taken.

✗ Branch 197 not taken.

✗ Branch 198 not taken.

✗ Branch 199 not taken.

✗ Branch 200 not taken.

✗ Branch 201 not taken.

✗ Branch 202 not taken.

✗ Branch 203 not taken.

✗ Branch 204 not taken.

✗ Branch 205 not taken.

✗ Branch 206 not taken.

✗ Branch 207 not taken.

✗ Branch 208 not taken.

✗ Branch 209 not taken.

✗ Branch 210 not taken.

✗ Branch 211 not taken.

✓ Branch 212 taken 354 times.

✗ Branch 213 not taken.

✓ Branch 214 taken 104 times.

✓ Branch 215 taken 250 times.

✓ Branch 216 taken 312 times.

✗ Branch 217 not taken.

✓ Branch 218 taken 101 times.

✓ Branch 219 taken 211 times.

✗ Branch 220 not taken.

✗ Branch 221 not taken.

✗ Branch 222 not taken.

✗ Branch 223 not taken.

189692

if ((cweight = contraction_find(wc, &chars_skipped))) return *cweight;

1546

}

1547

176945209

prev_char = wc;

1548

}

1549

1550

// For Japanese kana-sensitive collation.

1551

4/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 10033566 times.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 2420 times.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 449 times.

✗ Branch 23 not taken.

✓ Branch 24 taken 448 times.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

10036883

if (LEVELS_FOR_COMPARE == 4 && cs->coll_param == &ja_coll_param) {

1552

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 3345121 times.

✓ Branch 5 taken 6688445 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 779 times.

✓ Branch 13 taken 1641 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 134 times.

✓ Branch 23 taken 315 times.

✓ Branch 24 taken 133 times.

✓ Branch 25 taken 315 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

10036883

int quat_wt = handle_ja_common_quat_wt(wc);

1553

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 1392 times.

✓ Branch 5 taken 10032174 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 12 times.

✓ Branch 13 taken 2408 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 9 times.

✓ Branch 23 taken 440 times.

✓ Branch 24 taken 9 times.

✓ Branch 25 taken 439 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

10036883

if (quat_wt == -1)

1554

1422

continue;

1555

8/28

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 2892 times.

✓ Branch 5 taken 10029282 times.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✓ Branch 12 taken 71 times.

✓ Branch 13 taken 2337 times.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✓ Branch 22 taken 81 times.

✓ Branch 23 taken 359 times.

✓ Branch 24 taken 81 times.

✓ Branch 25 taken 358 times.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

10035461

else if (quat_wt)

1556

3125

return quat_wt;

1557

}

1558

/* Process single code point */

1559

38265827346

uint page = wc >> 8;

1560

38265827346

uint code = wc & 0xFF;

1561

1562

/* If weight page for wc does not exist, then calculate algorithmically */

1563

38265827346

const uint16 *wpage = uca->weights[page];

1564

35/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 9476838 times.

✓ Branch 17 taken 552444 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 289097553 times.

✓ Branch 21 taken 11208264 times.

✓ Branch 22 taken 9665721 times.

✓ Branch 23 taken 351943 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 6444000 times.

✓ Branch 27 taken 235748 times.

✓ Branch 28 taken 90209280 times.

✓ Branch 29 taken 3225166 times.

✓ Branch 30 taken 3228772 times.

✓ Branch 31 taken 52006306 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✓ Branch 49 taken 2337 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✓ Branch 53 taken 61605 times.

✗ Branch 54 not taken.

✓ Branch 55 taken 306 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✓ Branch 59 taken 28 times.

✗ Branch 60 not taken.

✓ Branch 61 taken 19161 times.

✓ Branch 62 taken 26 times.

✓ Branch 63 taken 223782822 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✓ Branch 89 taken 359 times.

✗ Branch 90 not taken.

✓ Branch 91 taken 358 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 142 times.

✓ Branch 95 taken 6500 times.

✓ Branch 96 taken 119 times.

✓ Branch 97 taken 6482 times.

✗ Branch 98 not taken.

✓ Branch 99 taken 57 times.

✓ Branch 100 taken 138 times.

✓ Branch 101 taken 14160 times.

✓ Branch 102 taken 117 times.

✓ Branch 103 taken 14171 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 8035 times.

✓ Branch 107 taken 18750557321 times.

✓ Branch 108 taken 10829 times.

✓ Branch 109 taken 18815637721 times.

✗ Branch 110 not taken.

✓ Branch 111 taken 2517 times.

38673968916

if (!wpage) return next_implicit(wc);

1565

1566

/* Calculate pointer to wc's weight, using page and offset */

1567

37857685776

wbeg = UCA900_WEIGHT_ADDR(wpage, weight_lv, code);

1568

37857685776

wbeg_stride = UCA900_DISTANCE_BETWEEN_WEIGHTS;

1569

555498

num_of_ce_left = UCA900_NUM_OF_CE(wpage, code);

1570

36/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 10503 times.

✓ Branch 17 taken 543333 times.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 273333 times.

✓ Branch 21 taken 10934931 times.

✓ Branch 22 taken 9204 times.

✓ Branch 23 taken 342739 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 6668 times.

✓ Branch 27 taken 229080 times.

✓ Branch 28 taken 114828 times.

✓ Branch 29 taken 3110338 times.

✓ Branch 30 taken 5235 times.

✓ Branch 31 taken 52001071 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 12 times.

✓ Branch 49 taken 2337 times.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✓ Branch 53 taken 61605 times.

✗ Branch 54 not taken.

✓ Branch 55 taken 306 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✓ Branch 59 taken 28 times.

✗ Branch 60 not taken.

✓ Branch 61 taken 19161 times.

✓ Branch 62 taken 440 times.

✓ Branch 63 taken 223782382 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✓ Branch 88 taken 9 times.

✓ Branch 89 taken 359 times.

✓ Branch 90 taken 9 times.

✓ Branch 91 taken 358 times.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 6 times.

✓ Branch 95 taken 6494 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 6482 times.

✗ Branch 98 not taken.

✓ Branch 99 taken 57 times.

✓ Branch 100 taken 1 times.

✓ Branch 101 taken 14159 times.

✗ Branch 102 not taken.

✓ Branch 103 taken 14171 times.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✓ Branch 106 taken 33593255 times.

✓ Branch 107 taken 18716964066 times.

✓ Branch 108 taken 58072832 times.

✓ Branch 109 taken 18757564889 times.

✗ Branch 110 not taken.

✓ Branch 111 taken 2517 times.

37857687198

} while (!wbeg[0]); /* Skip ignorable code points */

1571

1572

37765600863

uint16 rtn = *wbeg;

1573

37765600863

wbeg += wbeg_stride;

1574

37765600863

--num_of_ce_left;

1575

37765600863

return rtn;

1576

}

1577

1578

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1579

template <class T, class U>

1580

ALWAYS_INLINE void uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::for_each_weight(

1581

T func, U preaccept_data) {

1582

13/64

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 3344224 times.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✓ Branch 23 taken 3344224 times.

✓ Branch 24 taken 3344870 times.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 3344870 times.

✓ Branch 28 taken 43555055 times.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✓ Branch 31 taken 43555064 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✓ Branch 52 taken 51 times.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✓ Branch 55 taken 51 times.

✓ Branch 56 taken 14 times.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✓ Branch 59 taken 14 times.

✓ Branch 60 taken 151912063 times.

✗ Branch 61 not taken.

✓ Branch 62 taken 105 times.

✓ Branch 63 taken 151911958 times.

202156217

if (cs->tailoring || cs->mbminlen != 1 || cs->coll_param) {

1583

// Slower, generic path.

1584

int s_res;

1585

12/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✓ Branch 8 taken 20287995 times.

✓ Branch 9 taken 3342402 times.

✓ Branch 10 taken 603859076 times.

✓ Branch 11 taken 100270837 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✓ Branch 14 taken 185431022 times.

✓ Branch 15 taken 93599881 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✓ Branch 24 taken 2628 times.

✓ Branch 25 taken 38 times.

✓ Branch 26 taken 62001 times.

✓ Branch 27 taken 90 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✓ Branch 30 taken 19242 times.

✓ Branch 31 taken 84 times.

1204089331

while ((s_res = next()) >= 0) {

1586

11/48

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 20287995 times.

✗ Branch 17 not taken.

✓ Branch 18 taken 333 times.

✓ Branch 19 taken 20287662 times.

✓ Branch 20 taken 603859076 times.

✗ Branch 21 not taken.

✓ Branch 22 taken 498 times.

✓ Branch 23 taken 603858578 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✓ Branch 28 taken 185431022 times.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✓ Branch 31 taken 185431022 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✓ Branch 41 taken 2628 times.

✗ Branch 42 not taken.

✓ Branch 43 taken 62001 times.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✓ Branch 47 taken 19242 times.

809661964

if (!func(s_res, s_res == 0)) return;

1587

}

1588

197213332

return;

}

/*

Fast path. TODO: See if we can accept some character sets

1593

with tailorings.

1594

*/

1595

202156181

const uint16 *ascii_wpage =

1596

202156181

UCA900_WEIGHT_ADDR(uca->weights[0], /*level=*/weight_lv, /*subcode=*/0);

1597

1598

/*

1599

Precalculate the limit for the fast path below, taking care not to form

1600

pointers that are before sbeg, as those cannot be legally compared.

1601

(In particular, this catches the case of sbeg == send == nullptr.)

1602

*/

1603

12/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 3145988 times.

✓ Branch 11 taken 198236 times.

✓ Branch 12 taken 3146566 times.

✓ Branch 13 taken 198304 times.

✓ Branch 14 taken 34434167 times.

✓ Branch 15 taken 9120897 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 28 times.

✓ Branch 27 taken 23 times.

✓ Branch 28 taken 6 times.

✓ Branch 29 taken 8 times.

✓ Branch 30 taken 124853839 times.

✓ Branch 31 taken 27058119 times.

202156181

const uchar *send_local = (send - sbeg > 3) ? (send - 3) : sbeg;

1604

1605

301346658

for (;;) {

1606

/*

1607

We could have more weights left from the previous call to next()

1608

(if any) that we need to deal with.

1609

*/

1610

int s_res;

1611

11/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 3444795 times.

✓ Branch 11 taken 20040882 times.

✓ Branch 12 taken 3367697 times.

✓ Branch 13 taken 13362597 times.

✓ Branch 14 taken 3296700 times.

✓ Branch 15 taken 94403414 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 4 times.

✓ Branch 27 taken 459 times.

✓ Branch 28 taken 2 times.

✓ Branch 29 taken 56 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 375695147 times.

513612024

while ((s_res = more_weight()) >= 0) {

1612

11/48

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✓ Branch 20 taken 3444795 times.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✓ Branch 23 taken 3444795 times.

✓ Branch 24 taken 3367697 times.

✗ Branch 25 not taken.

✓ Branch 26 taken 24 times.

✓ Branch 27 taken 3367673 times.

✓ Branch 28 taken 3296694 times.

✗ Branch 29 not taken.

✓ Branch 30 taken 9 times.

✓ Branch 31 taken 3296685 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✓ Branch 43 taken 4 times.

✗ Branch 44 not taken.

✓ Branch 45 taken 2 times.

✗ Branch 46 not taken.

✓ Branch 47 taken 26 times.

10109069

if (!func(s_res, s_res == 0)) return;

}

/*

Loop in a simple fast path as long as we only have non-ignorable

1617

ASCII characters. These characters always have exactly a single weight

1618

and consist of only a single byte, so we can skip a lot of the checks

1619

we'd otherwise have to do.

1620

*/

1621

503502555

const uchar *sbeg_local = sbeg;

1622

33/96

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✓ Branch 30 taken 9441582 times.

✓ Branch 31 taken 10601019 times.

✓ Branch 32 taken 9441349 times.

✓ Branch 33 taken 233 times.

✓ Branch 34 taken 9441349 times.

✓ Branch 35 taken 10601252 times.

✓ Branch 36 taken 6296078 times.

✓ Branch 37 taken 7068789 times.

✓ Branch 38 taken 6295751 times.

✓ Branch 39 taken 327 times.

✓ Branch 40 taken 6295751 times.

✓ Branch 41 taken 7069116 times.

✓ Branch 42 taken 297507442 times.

✓ Branch 43 taken 88162919 times.

✓ Branch 44 taken 294430502 times.

✓ Branch 45 taken 3076873 times.

✓ Branch 46 taken 294430502 times.

✓ Branch 47 taken 91239792 times.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✗ Branch 62 not taken.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✓ Branch 78 taken 606 times.

✓ Branch 79 taken 402 times.

✓ Branch 80 taken 606 times.

✗ Branch 81 not taken.

✓ Branch 82 taken 606 times.

✓ Branch 83 taken 402 times.

✓ Branch 84 taken 348 times.

✓ Branch 85 taken 50 times.

✓ Branch 86 taken 348 times.

✗ Branch 87 not taken.

✓ Branch 88 taken 348 times.

✓ Branch 89 taken 50 times.

✓ Branch 90 taken 473259778 times.

✓ Branch 91 taken 375694112 times.

✓ Branch 92 taken 473260155 times.

✗ Branch 93 not taken.

✓ Branch 94 taken 473260265 times.

✓ Branch 95 taken 375693910 times.

1268033125

while (sbeg_local < send_local && preaccept_data(sizeof(uint32))) {

1623

/*

1624

Check if all four bytes are in the range 0x20..0x7e, inclusive.

1625

These have exactly one weight. Note that this unfortunately does not

1626

include tab and newline, which would otherwise be legal candidates.

1627

1628

See the FastOutOfRange unit test for verification that the bitfiddling

1629

trick used here is correct.

1630

*/

1631

uint32 four_bytes;

1632

783428821

memcpy(&four_bytes, sbeg_local, sizeof(four_bytes));

1633

12/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 1719 times.

✓ Branch 11 taken 9439630 times.

✓ Branch 12 taken 2270 times.

✓ Branch 13 taken 6293481 times.

✓ Branch 14 taken 291274850 times.

✓ Branch 15 taken 3155652 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 549 times.

✓ Branch 27 taken 57 times.

✓ Branch 28 taken 342 times.

✓ Branch 29 taken 6 times.

✓ Branch 30 taken 473259061 times.

✓ Branch 31 taken 1204 times.

783428821

if (((four_bytes + 0x01010101u) & 0x80808080) ||

1634

7/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 1719 times.

✗ Branch 11 not taken.

✓ Branch 12 taken 2270 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 291266947 times.

✓ Branch 15 taken 7903 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 549 times.

✗ Branch 27 not taken.

✓ Branch 28 taken 342 times.

✗ Branch 29 not taken.

✓ Branch 30 taken 473259130 times.

✗ Branch 31 not taken.

764538791

((four_bytes - 0x20202020u) & 0x80808080))

1635

break;

1636

764530957

const int s_res0 = ascii_wpage[sbeg_local[0]];

1637

764530957

const int s_res1 = ascii_wpage[sbeg_local[1]];

1638

764530957

const int s_res2 = ascii_wpage[sbeg_local[2]];

1639

764530957

const int s_res3 = ascii_wpage[sbeg_local[3]];

1640

6/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✓ Branch 11 taken 1719 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 2270 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 291266947 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 549 times.

✗ Branch 28 not taken.

✓ Branch 29 taken 342 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 473259130 times.

764530957

assert(s_res0 != 0);

1641

6/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✓ Branch 11 taken 1719 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 2270 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 291266947 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 549 times.

✗ Branch 28 not taken.

✓ Branch 29 taken 342 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 473259130 times.

764530957

assert(s_res1 != 0);

1642

6/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✓ Branch 11 taken 1719 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 2270 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 291266947 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 549 times.

✗ Branch 28 not taken.

✓ Branch 29 taken 342 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 473259130 times.

764530957

assert(s_res2 != 0);

1643

6/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✓ Branch 11 taken 1719 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 2270 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 291266947 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✓ Branch 27 taken 549 times.

✗ Branch 28 not taken.

✓ Branch 29 taken 342 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 473259130 times.

764530957

assert(s_res3 != 0);

1644

3/16

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 1719 times.

✗ Branch 11 not taken.

✓ Branch 12 taken 2270 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 291266947 times.

✗ Branch 15 not taken.

764530957

func(s_res0, /*is_level_separator=*/false);

1645

3/16

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 1719 times.

✗ Branch 11 not taken.

✓ Branch 12 taken 2270 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 291266947 times.

✗ Branch 15 not taken.

764529713

func(s_res1, /*is_level_separator=*/false);

1646

3/16

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 1719 times.

✗ Branch 11 not taken.

✓ Branch 12 taken 2270 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 291266947 times.

✗ Branch 15 not taken.

764529965

func(s_res2, /*is_level_separator=*/false);

1647

3/16

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 1719 times.

✗ Branch 11 not taken.

✓ Branch 12 taken 2270 times.

✗ Branch 13 not taken.

✓ Branch 14 taken 291266947 times.

✗ Branch 15 not taken.

764530246

func(s_res3, /*is_level_separator=*/false);

1648

764530570

sbeg_local += sizeof(uint32);

1649

}

1650

503502386

sbeg = sbeg_local;

1651

1652

// Do a single code point in the generic path.

1653

503500242

s_res = next_raw();

1654

10/32

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✓ Branch 10 taken 6688324 times.

✓ Branch 11 taken 13352558 times.

✓ Branch 12 taken 3344797 times.

✓ Branch 13 taken 10017800 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 94401530 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✓ Branch 26 taken 102 times.

✓ Branch 27 taken 357 times.

✓ Branch 28 taken 14 times.

✓ Branch 29 taken 42 times.

✗ Branch 30 not taken.

✓ Branch 31 taken 375694718 times.

503500242

if (s_res == 0) {

1655

// Level separator, so we have to update our page pointer.

1656

10033237

ascii_wpage += UCA900_DISTANCE_BETWEEN_LEVELS;

1657

}

1658

37/112

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✓ Branch 40 taken 16696784 times.

✓ Branch 41 taken 3344098 times.

✓ Branch 42 taken 16696784 times.

✗ Branch 43 not taken.

✓ Branch 44 taken 126 times.

✓ Branch 45 taken 16696658 times.

✓ Branch 46 taken 3344224 times.

✓ Branch 47 taken 16696658 times.

✓ Branch 48 taken 10017877 times.

✓ Branch 49 taken 3344720 times.

✓ Branch 50 taken 10017877 times.

✗ Branch 51 not taken.

✓ Branch 52 taken 126 times.

✓ Branch 53 taken 10017751 times.

✓ Branch 54 taken 3344846 times.

✓ Branch 55 taken 10017751 times.

✓ Branch 56 taken 55231120 times.

✓ Branch 57 taken 39170410 times.

✓ Branch 58 taken 55230804 times.

✗ Branch 59 not taken.

✓ Branch 60 taken 4381979 times.

✓ Branch 61 taken 50848825 times.

✓ Branch 62 taken 43552402 times.

✓ Branch 63 taken 50848812 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✓ Branch 94 taken 408 times.

✓ Branch 95 taken 51 times.

✗ Branch 96 not taken.

✓ Branch 97 taken 408 times.

✓ Branch 98 taken 51 times.

✓ Branch 99 taken 408 times.

✓ Branch 100 taken 42 times.

✓ Branch 101 taken 14 times.

✗ Branch 102 not taken.

✓ Branch 103 taken 42 times.

✓ Branch 104 taken 14 times.

✓ Branch 105 taken 42 times.

✓ Branch 106 taken 223782760 times.

✓ Branch 107 taken 151911958 times.

✓ Branch 108 taken 320 times.

✓ Branch 109 taken 223782703 times.

✓ Branch 110 taken 151911994 times.

✓ Branch 111 taken 223782987 times.

503500242

if (s_res < 0 || !func(s_res, s_res == 0)) return;

}

}

/**

Change a weight according to the reorder parameters.

1664

@param weight The weight to change

1665

@retval reordered weight

1666

*/

1667

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1668

199005436

uint16 uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::apply_reorder_param(

1669

uint16 weight) {

1670

/*

1671

Chinese collation's reordering is done in next_implicit() and

1672

modify_all_zh_pages(). See the comment on zh_reorder_param and

1673

change_zh_implicit().

1674

*/

1675

2/2

✓ Branch 0 taken 6502021 times.

✓ Branch 1 taken 93000697 times.

199005436

if (cs->coll_param == &zh_coll_param) return weight;

1676

186001394

const Reorder_param *param = cs->coll_param->reorder_param;

1677

4/4

✓ Branch 0 taken 92630596 times.

✓ Branch 1 taken 370101 times.

✓ Branch 2 taken 787196 times.

✓ Branch 3 taken 91843400 times.

186001394

if (weight >= START_WEIGHT_TO_REORDER && weight <= param->max_weight) {

1678

1/2

✓ Branch 0 taken 2182167 times.

✗ Branch 1 not taken.

4364334

for (int rec_ind = 0; rec_ind < param->wt_rec_num; ++rec_ind) {

1679

4364334

const Reorder_wt_rec *wt_rec = param->wt_rec + rec_ind;

1680

2/2

✓ Branch 0 taken 1569824 times.

✓ Branch 1 taken 612343 times.

4364334

if (weight >= wt_rec->old_wt_bdy.begin &&

1681

2/2

✓ Branch 0 taken 787196 times.

✓ Branch 1 taken 782628 times.

3139648

weight <= wt_rec->old_wt_bdy.end) {

1682

/*

1683

As commented in adjust_japanese_weight(), if this is a Japanese

1684

collation, for characters whose weight is between Latin and Kana

1685

group, and for the characters whose weight is between Kana and

1686

Han, we need to change their weight to be after all Han

1687

characters. We decide to give them the weights [FB86 0000 0000]

1688

[origin weight] to make sure the new weights are greater than

1689

the maximum implicit weight of Han characters. If this character's

1690

origin weight has more than one non-ignorable primary weight, for

1691

example, [AAAA 0020 0002][BBBB 0020 0002], both AAAA and BBBB need

1692

to be changed. The new weight should be:

1693

[FB86 0000 0000][AAAA 0020 0002][FB86 0000 0000][BBBB 0020 0002].

1694

*/

1695

4/4

✓ Branch 0 taken 635782 times.

✓ Branch 1 taken 151414 times.

✓ Branch 2 taken 604098 times.

✓ Branch 3 taken 31684 times.

1574392

if (param == &ja_reorder_param && wt_rec->new_wt_bdy.begin == 0) {

1696

1208196

return_origin_weight = !return_origin_weight;

1697

2/2

✓ Branch 0 taken 302028 times.

✓ Branch 1 taken 302070 times.

1208196

if (return_origin_weight) break;

1698

1699

/*

1700

We didn't consume the weight; rewind the iterator, so we will

1701

get another call where we can output it.

1702

*/

1703

604140

wbeg -= wbeg_stride;

1704

604140

++num_of_ce_left;

1705

604140

return 0xFB86;

1706

}

1707

1708

// Regular (non-Japanese-specific) reordering.

1709

366196

return weight - wt_rec->old_wt_bdy.begin + wt_rec->new_wt_bdy.begin;

}

}

}

185031058

return weight;

1714

}

1715

1716

// See Unicode TR35 section 3.14.1.

1717

3415893

static bool is_tertiary_weight_upper_case(uint16 weight) {

1718

10/10

✓ Branch 0 taken 17050 times.

✓ Branch 1 taken 3398843 times.

✓ Branch 2 taken 8787 times.

✓ Branch 3 taken 8263 times.

✓ Branch 4 taken 3407405 times.

✓ Branch 5 taken 225 times.

✓ Branch 6 taken 3407168 times.

✓ Branch 7 taken 237 times.

✓ Branch 8 taken 3406829 times.

✓ Branch 9 taken 339 times.

3415893

if ((weight >= 0x08 && weight <= 0x0C) || weight == 0x0E || weight == 0x11 ||

1719

2/2

✓ Branch 0 taken 684 times.

✓ Branch 1 taken 3406145 times.

3406829

weight == 0x12 || weight == 0x1D)

1720

9748

return true;

1721

3406145

return false;

1722

}

1723

1724

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1725

26889090

uint16 uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::apply_case_first(

1726

uint16 weight) {

1727

/*

1728

We only apply case weight change here when the character is not tailored.

1729

Tailored character's case weight has been changed in

1730

my_char_weight_put_900().

1731

We have only 1 collation (Danish) needs to implement [caseFirst upper].

1732

*/

1733

5/6

✓ Branch 0 taken 13444545 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 3410862 times.

✓ Branch 3 taken 10033683 times.

✓ Branch 4 taken 3410555 times.

✓ Branch 5 taken 307 times.

26889090

if (cs->coll_param->case_first == CASE_FIRST_UPPER && weight_lv == 2 &&

1734

weight < 0x20) {

1735

2/2

✓ Branch 0 taken 7079 times.

✓ Branch 1 taken 3403476 times.

6821110

if (is_tertiary_weight_upper_case(weight))

1736

14158

weight |= CASE_FIRST_UPPER_MASK;

1737

else

1738

6806952

weight |= CASE_FIRST_LOWER_MASK;

1739

}

1740

26889090

return weight;

1741

}

1742

1743

template <class Mb_wc, int LEVELS_FOR_COMPARE>

1744

ALWAYS_INLINE int uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE>::next() {

1745

39218166089

int res = next_raw();

1746

39218166089

Coll_param *param = cs->coll_param;

1747

56/160

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 13602000 times.

✓ Branch 17 taken 10028397 times.

✓ Branch 18 taken 13602000 times.

✗ Branch 19 not taken.

✓ Branch 20 taken 403316844 times.

✓ Branch 21 taken 300813069 times.

✓ Branch 22 taken 161350512 times.

✓ Branch 23 taken 241966332 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✓ Branch 28 taken 185431022 times.

✓ Branch 29 taken 93599881 times.

✓ Branch 30 taken 39732257 times.

✓ Branch 31 taken 145698765 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 2520 times.

✓ Branch 49 taken 146 times.

✓ Branch 50 taken 2520 times.

✗ Branch 51 not taken.

✓ Branch 52 taken 61821 times.

✓ Branch 53 taken 270 times.

✓ Branch 54 taken 24732 times.

✓ Branch 55 taken 37089 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✓ Branch 60 taken 19242 times.

✓ Branch 61 taken 84 times.

✓ Branch 62 taken 4122 times.

✓ Branch 63 taken 15120 times.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✓ Branch 112 taken 564 times.

✓ Branch 113 taken 162 times.

✓ Branch 114 taken 564 times.

✗ Branch 115 not taken.

✓ Branch 116 taken 563 times.

✓ Branch 117 taken 163 times.

✓ Branch 118 taken 563 times.

✗ Branch 119 not taken.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✓ Branch 124 taken 6933 times.

✓ Branch 125 taken 2119 times.

✓ Branch 126 taken 472 times.

✓ Branch 127 taken 6461 times.

✓ Branch 128 taken 6880 times.

✓ Branch 129 taken 2172 times.

✓ Branch 130 taken 456 times.

✓ Branch 131 taken 6424 times.

✓ Branch 132 taken 57 times.

✓ Branch 133 taken 39 times.

✗ Branch 134 not taken.

✓ Branch 135 taken 57 times.

✓ Branch 136 taken 14390 times.

✓ Branch 137 taken 1816 times.

✗ Branch 138 not taken.

✓ Branch 139 taken 14390 times.

✓ Branch 140 taken 14381 times.

✓ Branch 141 taken 1825 times.

✗ Branch 142 not taken.

✓ Branch 143 taken 14381 times.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✓ Branch 148 taken 18739260764 times.

✓ Branch 149 taken 346531268 times.

✓ Branch 150 taken 783 times.

✓ Branch 151 taken 18739259981 times.

✓ Branch 152 taken 18752629280 times.

✓ Branch 153 taken 372813301 times.

✓ Branch 154 taken 800 times.

✓ Branch 155 taken 18752628480 times.

✓ Branch 156 taken 2517 times.

✓ Branch 157 taken 1599 times.

✗ Branch 158 not taken.

✓ Branch 159 taken 2517 times.

39218166089

if (res > 0 && param) {

1748

/* Reorder weight change only on primary level. */

1749

36/160

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✗ Branch 9 not taken.

✗ Branch 10 not taken.

✗ Branch 11 not taken.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✗ Branch 15 not taken.

✓ Branch 16 taken 13602000 times.

✗ Branch 17 not taken.

✓ Branch 18 taken 6764423 times.

✓ Branch 19 taken 6837577 times.

✓ Branch 20 taken 107580689 times.

✓ Branch 21 taken 53769823 times.

✓ Branch 22 taken 52993719 times.

✓ Branch 23 taken 54586970 times.

✗ Branch 24 not taken.

✗ Branch 25 not taken.

✗ Branch 26 not taken.

✗ Branch 27 not taken.

✓ Branch 28 taken 39732257 times.

✗ Branch 29 not taken.

✓ Branch 30 taken 39732257 times.

✗ Branch 31 not taken.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✓ Branch 48 taken 2520 times.

✗ Branch 49 not taken.

✓ Branch 50 taken 793 times.

✓ Branch 51 taken 1727 times.

✓ Branch 52 taken 16488 times.

✓ Branch 53 taken 8244 times.

✓ Branch 54 taken 5496 times.

✓ Branch 55 taken 10992 times.

✗ Branch 56 not taken.

✗ Branch 57 not taken.

✗ Branch 58 not taken.

✗ Branch 59 not taken.

✓ Branch 60 taken 4122 times.

✗ Branch 61 not taken.

✓ Branch 62 taken 4122 times.

✗ Branch 63 not taken.

✗ Branch 64 not taken.

✗ Branch 65 not taken.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✗ Branch 75 not taken.

✗ Branch 76 not taken.

✗ Branch 77 not taken.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

✗ Branch 80 not taken.

✗ Branch 81 not taken.

✗ Branch 82 not taken.

✗ Branch 83 not taken.

✗ Branch 84 not taken.

✗ Branch 85 not taken.

✗ Branch 86 not taken.

✗ Branch 87 not taken.

✗ Branch 88 not taken.

✗ Branch 89 not taken.

✗ Branch 90 not taken.

✗ Branch 91 not taken.

✗ Branch 92 not taken.

✗ Branch 93 not taken.

✗ Branch 94 not taken.

✗ Branch 95 not taken.

✗ Branch 96 not taken.

✗ Branch 97 not taken.

✗ Branch 98 not taken.

✗ Branch 99 not taken.

✗ Branch 100 not taken.

✗ Branch 101 not taken.

✗ Branch 102 not taken.

✗ Branch 103 not taken.

✗ Branch 104 not taken.

✗ Branch 105 not taken.

✗ Branch 106 not taken.

✗ Branch 107 not taken.

✗ Branch 108 not taken.

✗ Branch 109 not taken.

✗ Branch 110 not taken.

✗ Branch 111 not taken.

✓ Branch 112 taken 564 times.

✗ Branch 113 not taken.

✓ Branch 114 taken 149 times.

✓ Branch 115 taken 415 times.

✓ Branch 116 taken 563 times.

✗ Branch 117 not taken.

✓ Branch 118 taken 148 times.

✓ Branch 119 taken 415 times.

✗ Branch 120 not taken.

✗ Branch 121 not taken.

✗ Branch 122 not taken.

✗ Branch 123 not taken.

✓ Branch 124 taken 41 times.

✓ Branch 125 taken 431 times.

✓ Branch 126 taken 14 times.

✓ Branch 127 taken 27 times.

✓ Branch 128 taken 45 times.

✓ Branch 129 taken 411 times.

✓ Branch 130 taken 14 times.

✓ Branch 131 taken 31 times.

✗ Branch 132 not taken.

✗ Branch 133 not taken.

✗ Branch 134 not taken.

✗ Branch 135 not taken.

✗ Branch 136 not taken.

✗ Branch 137 not taken.

✗ Branch 138 not taken.

✗ Branch 139 not taken.

✗ Branch 140 not taken.

✗ Branch 141 not taken.

✗ Branch 142 not taken.

✗ Branch 143 not taken.

✗ Branch 144 not taken.

✗ Branch 145 not taken.

✗ Branch 146 not taken.

✗ Branch 147 not taken.

✓ Branch 148 taken 783 times.

✗ Branch 149 not taken.

✓ Branch 150 taken 783 times.

✗ Branch 151 not taken.

✓ Branch 152 taken 800 times.

✗ Branch 153 not taken.

✓ Branch 154 taken 800 times.

✗ Branch 155 not taken.

✗ Branch 156 not taken.

✗ Branch 157 not taken.

✗ Branch 158 not taken.

✗ Branch 159 not taken.

214719781

if (param->reorder_param && weight_lv == 0) res = apply_reorder_param(res);

1750

16/80

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✓ Branch 9 taken 13602000 times.

✓ Branch 10 taken 13442191 times.

✓ Branch 11 taken 147908321 times.

✗ Branch 12 not taken.

✗ Branch 13 not taken.

✗ Branch 14 not taken.

✓ Branch 15 taken 39732257 times.

✗ Branch 16 not taken.

✗ Branch 17 not taken.

✗ Branch 18 not taken.

✗ Branch 19 not taken.

✗ Branch 20 not taken.

✗ Branch 21 not taken.

✗ Branch 22 not taken.

✗ Branch 23 not taken.

✗ Branch 24 not taken.

✓ Branch 25 taken 2520 times.

✓ Branch 26 taken 2061 times.

✓ Branch 27 taken 22671 times.

✗ Branch 28 not taken.

✗ Branch 29 not taken.

✗ Branch 30 not taken.

✓ Branch 31 taken 4122 times.

✗ Branch 32 not taken.

✗ Branch 33 not taken.

✗ Branch 34 not taken.

✗ Branch 35 not taken.

✗ Branch 36 not taken.

✗ Branch 37 not taken.

✗ Branch 38 not taken.

✗ Branch 39 not taken.

✗ Branch 40 not taken.

✗ Branch 41 not taken.

✗ Branch 42 not taken.

✗ Branch 43 not taken.

✗ Branch 44 not taken.

✗ Branch 45 not taken.

✗ Branch 46 not taken.

✗ Branch 47 not taken.

✗ Branch 48 not taken.

✗ Branch 49 not taken.

✗ Branch 50 not taken.

✗ Branch 51 not taken.

✗ Branch 52 not taken.

✗ Branch 53 not taken.

✗ Branch 54 not taken.

✗ Branch 55 not taken.

✗ Branch 56 not taken.

✓ Branch 57 taken 564 times.

✗ Branch 58 not taken.

✓ Branch 59 taken 563 times.

✗ Branch 60 not taken.

✗ Branch 61 not taken.

✓ Branch 62 taken 151 times.

✓ Branch 63 taken 321 times.

✓ Branch 64 taken 142 times.

✓ Branch 65 taken 314 times.

✗ Branch 66 not taken.

✗ Branch 67 not taken.

✗ Branch 68 not taken.

✗ Branch 69 not taken.

✗ Branch 70 not taken.

✗ Branch 71 not taken.

✗ Branch 72 not taken.

✗ Branch 73 not taken.

✗ Branch 74 not taken.

✓ Branch 75 taken 783 times.

✗ Branch 76 not taken.

✓ Branch 77 taken 800 times.

✗ Branch 78 not taken.

✗ Branch 79 not taken.

214719781

if (param->case_first != CASE_FIRST_OFF) res = apply_case_first(res);

1751

}

1752

39218166089

return res;

}

/*

Compares two strings according to the collation

SYNOPSIS:

my_strnncoll_uca()

cs Character set information

1761

s First string

1762

slen First string length

1763

t Second string

1764

tlen Second string length

1765

1766

NOTES:

1767

Initializes two weight scanners and gets weights

1768

corresponding to two strings in a loop. If weights are not

1769

the same at some step then returns their difference.

1770

1771

In the while() comparison these situations are possible:

1772

1. (s_res>0) and (t_res>0) and (s_res == t_res)

1773

Weights are the same so far, continue comparison

1774

2. (s_res>0) and (t_res>0) and (s_res!=t_res)

1775

A difference has been found, return.

1776

3. (s_res>0) and (t_res<0)

1777

We have reached the end of the second string, or found

1778

an illegal multibyte sequence in the second string.

1779

Return a positive number, i.e. the first string is bigger.

1780

4. (s_res<0) and (t_res>0)

1781

We have reached the end of the first string, or found

1782

an illegal multibyte sequence in the first string.

1783

Return a negative number, i.e. the second string is bigger.

1784

5. (s_res<0) and (t_res<0)

1785

Both scanners returned -1. It means we have riched

1786

the end-of-string of illegal-sequence in both strings

1787

at the same time. Return 0, strings are equal.

1788

1789

RETURN

1790

Difference between two strings, according to the collation:

1791

0 - means strings are equal

1792

negative number - means the first string is smaller

1793

positive number - means the first string is bigger

1794

*/

1795

1796

template <class Scanner, int LEVELS_FOR_COMPARE, class Mb_wc>

1797

10423188002

static int my_strnncoll_uca(const CHARSET_INFO *cs, const Mb_wc mb_wc,

1798

const uchar *s, size_t slen, const uchar *t,

1799

size_t tlen, bool t_is_prefix) {

1800

10423188002

Scanner sscanner(mb_wc, cs, s, slen);

1801

10423073348

Scanner tscanner(mb_wc, cs, t, tlen);

1802

10422907406

int s_res = 0;

1803

10422907406

int t_res = 0;

1804

1805

/*

1806

We compare 2 strings in same level first. If only string A's scanner

1807

has gone to next level, which means another string, B's weight of

1808

current level is longer than A's. We'll compare B's remaining weights

1809

with space.

1810

*/

1811

1/2

✓ Branch 0 taken 5211497328 times.

✗ Branch 1 not taken.

10422913132

for (uint current_lv = 0; current_lv < LEVELS_FOR_COMPARE; ++current_lv) {

1812

/* Run the scanners until one of them runs out of current lv */

1813

do {

1814

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 182 times.

38171636396

s_res = sscanner.next();

1815

38250937494

t_res = tscanner.next();

1816

4/4

✓ Branch 0 taken 13973876528 times.

✓ Branch 1 taken 403196250 times.

✓ Branch 2 taken 13863831101 times.

✓ Branch 3 taken 90007467 times.

56661822692

} while (s_res == t_res && s_res >= 0 &&

1817

5/6

✓ Branch 0 taken 14377072778 times.

✓ Branch 1 taken 4748395969 times.

✓ Branch 2 taken 13953403909 times.

✗ Branch 3 not taken.

✓ Branch 4 taken 13953360344 times.

✓ Branch 5 taken 5211930012 times.

95032479324

sscanner.get_weight_level() == current_lv &&

1818

27727662202

tscanner.get_weight_level() == current_lv);

1819

1820

/*

1821

Two scanners run to next level at same time, or we found a difference,

1822

or we found an error.

1823

*/

1824

2/2

✓ Branch 0 taken 5207174308 times.

✓ Branch 1 taken 4710188 times.

10423860024

if (sscanner.get_weight_level() == tscanner.get_weight_level()) {

1825

4/4

✓ Branch 0 taken 366010689 times.

✓ Branch 1 taken 4841163619 times.

✓ Branch 2 taken 2833 times.

✓ Branch 3 taken 366007856 times.

10414348616

if (s_res == t_res && s_res >= 0) continue;

1826

10414342950

break; // Error or inequality found, end.

1827

}

1828

1829

2/2

✓ Branch 0 taken 1852950 times.

✓ Branch 1 taken 2857238 times.

9420376

if (tscanner.get_weight_level() > current_lv) {

1830

// t ran out of weights on this level, and s didn't.

1831

2/2

✓ Branch 0 taken 1638 times.

✓ Branch 1 taken 1851312 times.

3705900

if (t_is_prefix) {

1832

// Consume the rest of the weights from s.

1833

do {

1834

8424

s_res = sscanner.next();

1835

6/6

✓ Branch 0 taken 2604 times.

✓ Branch 1 taken 1608 times.

✓ Branch 2 taken 2574 times.

✓ Branch 3 taken 30 times.

✓ Branch 4 taken 2574 times.

✓ Branch 5 taken 1638 times.

8424

} while (s_res >= 0 && sscanner.get_weight_level() == current_lv);

1836

1837

2/2

✓ Branch 0 taken 1608 times.

✓ Branch 1 taken 30 times.

3276

if (s_res < 0) break; // Error found, end.

1838

1839

// s is now also on the next level. Continue comparison.

1840

60

continue;

1841

} else {

1842

// s is longer than t (and t_prefix isn't set).

1843

3702624

return 1;

}

}

1/2

✓ Branch 0 taken 2857238 times.

✗ Branch 1 not taken.

5714476

if (sscanner.get_weight_level() > current_lv) {

1848

// s ran out of weights on this level, and t didn't.

1849

5714476

return -1;

1850

}

1851

1852

✗

break;

1853

}

1854

1855

10414264642

return (s_res - t_res);

1856

}

1857

1858

974136

static inline int my_space_weight(const CHARSET_INFO *cs) /* W3-TODO */

1859

{

1860

2/4

✓ Branch 0 taken 974136 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 974136 times.

974136

if (cs->uca && cs->uca->version == UCA_V900)

1861

✗

return UCA900_WEIGHT(cs->uca->weights[0], /*weight_lv=*/0, 0x20);

1862

else

1863

974136

return cs->uca->weights[0][0x20 * cs->uca->lengths[0]];

}

/**

Helper function:

Find address of weights of the given code point.

1869

1870

@param uca Pointer to UCA data

1871

@param wc character Unicode code point

1872

1873

@return Weight array

1874

@retval pointer to weight array for the given code point,

1875

or nullptr if this page does not have implicit weights.

1876

*/

1877

1878

1018732

static inline uint16 *my_char_weight_addr(MY_UCA_INFO *uca, my_wc_t wc) {

1879

uint page, ofst;

1880

1/2

✓ Branch 0 taken 1018732 times.

✗ Branch 1 not taken.

2037464

return wc > uca->maxchar ? nullptr

1881

1018732

: (uca->weights[page = (wc >> 8)]

1882

1/2

✓ Branch 0 taken 1018732 times.

✗ Branch 1 not taken.

1018732

? uca->weights[page] + (ofst = (wc & 0xFF)) *

1883

1018732

uca->lengths[page]

1884

1018732

: nullptr);

}

/**

Helper function:

Find address of weights of the given code point, for UCA 9.0.0 format.

1890

1891

@param uca Pointer to UCA data

1892

@param wc character Unicode code point

1893

1894

@return Weight array

1895

@retval pointer to weight array for the given code point,

1896

or nullptr if this page does not have implicit weights.

1897

*/

1898

1899

296798798

static inline uint16 *my_char_weight_addr_900(MY_UCA_INFO *uca, my_wc_t wc) {

1900

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 296798798 times.

296798798

if (wc > uca->maxchar) return nullptr;

1901

1902

296798798

uint page = wc >> 8;

1903

296798798

uint ofst = wc & 0xFF;

1904

296798798

uint16 *weights = uca->weights[page];

1905

2/2

✓ Branch 0 taken 296795184 times.

✓ Branch 1 taken 3614 times.

296798798

if (weights)

1906

296795184

return UCA900_WEIGHT_ADDR(weights, /*level=*/0, ofst);

1907

else

1908

3614

return nullptr;

}

/*

Compares two strings according to the collation,

1913

ignoring trailing spaces.

SYNOPSIS:

my_strnncollsp_uca()

cs Character set information

1918

s First string

1919

slen First string length

1920

t Second string

1921

tlen Second string length

1922

1923

NOTES:

1924

Works exactly the same with my_strnncoll_uca(),

1925

but ignores trailing spaces.

1926

1927

In the while() comparison these situations are possible:

1928

1. (s_res>0) and (t_res>0) and (s_res == t_res)

1929

Weights are the same so far, continue comparison

1930

2. (s_res>0) and (t_res>0) and (s_res!=t_res)

1931

A difference has been found, return.

1932

3. (s_res>0) and (t_res<0)

1933

We have reached the end of the second string, or found

1934

an illegal multibyte sequence in the second string.

1935

Compare the first string to an infinite array of

1936

space characters until difference is found, or until

1937

the end of the first string.

1938

4. (s_res<0) and (t_res>0)

1939

We have reached the end of the first string, or found

1940

an illegal multibyte sequence in the first string.

1941

Compare the second string to an infinite array of

1942

space characters until difference is found or until

1943

the end of the second steing.

1944

5. (s_res<0) and (t_res<0)

1945

Both scanners returned -1. It means we have riched

1946

the end-of-string of illegal-sequence in both strings

1947

at the same time. Return 0, strings are equal.

1948

1949

RETURN

1950

Difference between two strings, according to the collation:

1951

0 - means strings are equal

1952

negative number - means the first string is smaller

1953

positive number - means the first string is bigger

1954

*/

1955

1956

template <class Mb_wc>

1957

553356

static int my_strnncollsp_uca(const CHARSET_INFO *cs, Mb_wc mb_wc,

1958

const uchar *s, size_t slen, const uchar *t,

size_t tlen) {

int s_res, t_res;

553356

uca_scanner_any<Mb_wc> sscanner(mb_wc, cs, s, slen);

1963

553366

uca_scanner_any<Mb_wc> tscanner(mb_wc, cs, t, tlen);

1964

1965

do {

1966

2/2

✓ Branch 0 taken 105108 times.

✓ Branch 1 taken 7728108 times.

15666432

s_res = sscanner.next();

1967

15667062

t_res = tscanner.next();

1968

4/4

✓ Branch 0 taken 7805855 times.

✓ Branch 1 taken 27676 times.

✓ Branch 2 taken 7556839 times.

✓ Branch 3 taken 249016 times.

15667062

} while (s_res == t_res && s_res > 0);

1969

1970

4/4

✓ Branch 0 taken 21618 times.

✓ Branch 1 taken 255074 times.

✓ Branch 2 taken 438 times.

✓ Branch 3 taken 21180 times.

553384

if (s_res > 0 && t_res < 0) {

1971

/* Calculate weight for SPACE character */

1972

876

t_res = my_space_weight(cs);

1973

1974

/* compare the first string to spaces */

1975

do {

1976

2/2

✓ Branch 0 taken 201 times.

✓ Branch 1 taken 376 times.

1154

if (s_res != t_res) return (s_res - t_res);

1977

752

s_res = sscanner.next();

1978

2/2

✓ Branch 0 taken 139 times.

✓ Branch 1 taken 237 times.

752

} while (s_res > 0);

1979

474

return 0;

1980

}

1981

1982

4/4

✓ Branch 0 taken 255075 times.

✓ Branch 1 taken 21179 times.

✓ Branch 2 taken 6015 times.

✓ Branch 3 taken 249060 times.

552508

if (s_res < 0 && t_res > 0) {

1983

/* Calculate weight for SPACE character */

1984

12030

s_res = my_space_weight(cs);

1985

1986

/* compare the second string to spaces */

1987

do {

1988

2/2

✓ Branch 0 taken 5577 times.

✓ Branch 1 taken 1448 times.

14050

if (s_res != t_res) return (s_res - t_res);

1989

2896

t_res = tscanner.next();

1990

2/2

✓ Branch 0 taken 1010 times.

✓ Branch 1 taken 438 times.

2896

} while (t_res > 0);

1991

876

return 0;

1992

}

1993

1994

540478

return (s_res - t_res);

}

/*

Calculates hash value for the given string,

1999

according to the collation, and ignoring trailing spaces.

SYNOPSIS:

my_hash_sort_uca()

cs Character set information

2004

s String

2005

slen String's length

2006

n1 First hash parameter

2007

n2 Second hash parameter

2008

2009

NOTES:

2010

Scans consequently weights and updates

2011

hash parameters n1 and n2. In a case insensitive collation,

2012

upper and lower case of the same letter will return the same

2013

weight sequence, and thus will produce the same hash values

in n1 and n2.

RETURN

N/A

*/

template <class Mb_wc>

2021

5568

static void my_hash_sort_uca(const CHARSET_INFO *cs, Mb_wc mb_wc,

2022

const uchar *s, size_t slen, uint64 *n1,

uint64 *n2) {

int s_res;

uint64 tmp1;

uint64 tmp2;

1/2

✓ Branch 0 taken 2784 times.

✗ Branch 1 not taken.

5568

slen = cs->cset->lengthsp(cs, pointer_cast<const char *>(s), slen);

2029

5568

uca_scanner_any<Mb_wc> scanner(mb_wc, cs, s, slen);

2030

2031

5568

tmp1 = *n1;

2032

5568

tmp2 = *n2;

2033

2034

2/2

✓ Branch 0 taken 126188 times.

✓ Branch 1 taken 2784 times.

263512

while ((s_res = scanner.next()) > 0) {

2035

252376

tmp1 ^= (((tmp1 & 63) + tmp2) * (s_res >> 8)) + (tmp1 << 8);

2036

252376

tmp2 += 3;

2037

252376

tmp1 ^= (((tmp1 & 63) + tmp2) * (s_res & 0xFF)) + (tmp1 << 8);

2038

252376

tmp2 += 3;

2039

}

2040

2041

5568

*n1 = tmp1;

2042

5568

*n2 = tmp2;

2043

5568

}

2044

2045

/*

2046

For the given string creates its "binary image", suitable

2047

to be used in binary comparison, i.e. in memcmp().

SYNOPSIS:

my_strnxfrm_uca()

cs Character set information

2052

dst Where to write the image

2053

dstlen Space available for the image, in bytes

2054

src The source string

2055

srclen Length of the source string, in bytes

2056

2057

NOTES:

2058

In a loop, scans weights from the source string and writes

2059

them into the binary image. In a case insensitive collation,

2060

upper and lower cases of the same letter will produce the

2061

same image subsequences. When we have reached the end-of-string

2062

or found an illegal multibyte sequence, the loop stops.

2063

2064

It is impossible to restore the original string using its

2065

binary image.

2066

2067

Binary images are used for bulk comparison purposes,

2068

e.g. in ORDER BY, when it is more efficient to create

2069

a binary image and use it instead of weight scanner

2070

for the original strings for every comparison.

2071

2072

RETURN

2073

Number of bytes that have been written into the binary image.

2074

*/

2075

2076

template <class Mb_wc>

2077

507239098

static size_t my_strnxfrm_uca(const CHARSET_INFO *cs, Mb_wc mb_wc, uchar *dst,

2078

size_t dstlen, uint num_codepoints,

2079

const uchar *src, size_t srclen, uint flags) {

2080

507239098

uchar *d0 = dst;

2081

507239098

uchar *de = dst + dstlen;

2082

int s_res;

2083

507239098

uca_scanner_any<Mb_wc> scanner(mb_wc, cs, src, srclen);

2084

2085

6/6

✓ Branch 0 taken 540461295 times.

✓ Branch 1 taken 248 times.

✓ Branch 2 taken 286841994 times.

✓ Branch 3 taken 253619301 times.

✓ Branch 4 taken 286841994 times.

✓ Branch 5 taken 253619549 times.

2161845676

while (dst < de && (s_res = scanner.next()) > 0) {

2086

573683988

*dst++ = s_res >> 8;

2087

1/2

✓ Branch 0 taken 286841994 times.

✗ Branch 1 not taken.

573683988

if (dst < de) *dst++ = s_res & 0xFF;

2088

}

2089

2090

2/2

✓ Branch 0 taken 253619301 times.

✓ Branch 1 taken 248 times.

507239098

if (dst < de) {

/*

PAD SPACE behavior.

We still have space left in the output buffer, which must mean

2095

that the scanner is at the end of the last level. Find out

2096

how many weights we wrote per level, and add any remaining

2097

spaces we need to get us up to the requested total.

2098

*/

2099

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 253619301 times.

507238602

assert(num_codepoints >= scanner.get_char_index());

2100

507238602

num_codepoints -= scanner.get_char_index();

2101

2102

2/2

✓ Branch 0 taken 627887 times.

✓ Branch 1 taken 252991414 times.

507238602

if (num_codepoints) {

2103

1255774

uint space_count = std::min<uint>((de - dst) / 2, num_codepoints);

2104

1255774

s_res = my_space_weight(cs);

2105

2/2

✓ Branch 0 taken 1783514 times.

✓ Branch 1 taken 627887 times.

4822802

for (; space_count; space_count--) {

2106

7134056

dst = store16be(dst, s_res);

}

}

}

4/4

✓ Branch 0 taken 340005 times.

✓ Branch 1 taken 253279544 times.

✓ Branch 2 taken 339796 times.

✓ Branch 3 taken 209 times.

507239098

if ((flags & MY_STRXFRM_PAD_TO_MAXLEN) && dst < de) {

2111

679592

s_res = my_space_weight(cs);

2112

2/2

✓ Branch 0 taken 23797171 times.

✓ Branch 1 taken 339796 times.

48273934

for (; dst < de;) {

2113

47594342

*dst++ = s_res >> 8;

2114

1/2

✓ Branch 0 taken 23797171 times.

✗ Branch 1 not taken.

47594342

if (dst < de) *dst++ = s_res & 0xFF;

2115

}

2116

}

2117

507239098

return dst - d0;

2118

}

2119

2120

148148069

static int my_uca_charcmp_900(const CHARSET_INFO *cs, my_wc_t wc1,

2121

my_wc_t wc2) {

2122

148148069

uint16 *weight1_ptr = my_char_weight_addr_900(cs->uca, wc1); /* W3-TODO */

2123

148148069

uint16 *weight2_ptr = my_char_weight_addr_900(cs->uca, wc2);

2124

2125

/* Check if some of the characters does not have implicit weights */

2126

4/4

✓ Branch 0 taken 148146934 times.

✓ Branch 1 taken 1135 times.

✓ Branch 2 taken 1812 times.

✓ Branch 3 taken 148145122 times.

148148069

if (!weight1_ptr || !weight2_ptr) return wc1 != wc2;

2127

2128

5/6

✓ Branch 0 taken 148144905 times.

✓ Branch 1 taken 217 times.

✓ Branch 2 taken 148144905 times.

✗ Branch 3 not taken.

✓ Branch 4 taken 146229511 times.

✓ Branch 5 taken 1915394 times.

148145122

if (weight1_ptr[0] && weight2_ptr[0] && weight1_ptr[0] != weight2_ptr[0])

2129

146229511

return 1;

2130

2131

/* Thoroughly compare all weights */

2132

1915611

size_t length1 = weight1_ptr[-UCA900_DISTANCE_BETWEEN_LEVELS];

2133

1915611

size_t length2 = weight2_ptr[-UCA900_DISTANCE_BETWEEN_LEVELS];

2134

2135

2/2

✓ Branch 0 taken 1915708 times.

✓ Branch 1 taken 1915364 times.

3831072

for (int level = 0; level < cs->levels_for_compare; ++level) {

2136

1915708

size_t wt_ind1 = 0;

2137

1915708

size_t wt_ind2 = 0;

2138

1915708

uint16 *weight1 = weight1_ptr + level * UCA900_DISTANCE_BETWEEN_LEVELS;

2139

1915708

uint16 *weight2 = weight2_ptr + level * UCA900_DISTANCE_BETWEEN_LEVELS;

2140

4/4

✓ Branch 0 taken 1915856 times.

✓ Branch 1 taken 1915385 times.

✓ Branch 2 taken 1915814 times.

✓ Branch 3 taken 42 times.

3831241

while (wt_ind1 < length1 && wt_ind2 < length2) {

2141

// Zero weight is ignorable.

2142

4/4

✓ Branch 0 taken 1915814 times.

✓ Branch 1 taken 272 times.

✓ Branch 2 taken 272 times.

✓ Branch 3 taken 1915542 times.

1916086

for (; wt_ind1 < length1 && !*weight1; wt_ind1++)

2143

272

weight1 += UCA900_DISTANCE_BETWEEN_WEIGHTS;

2144

2/2

✓ Branch 0 taken 272 times.

✓ Branch 1 taken 1915542 times.

1915814

if (wt_ind1 == length1) break;

2145

2/4

✓ Branch 0 taken 1915542 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 1915542 times.

1915542

for (; wt_ind2 < length2 && !*weight2; wt_ind2++)

2146

✗

weight2 += UCA900_DISTANCE_BETWEEN_WEIGHTS;

2147

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1915542 times.

1915542

if (wt_ind2 == length2) break;

2148

2149

// Check if these two non-ignorable weights are equal.

2150

2/2

✓ Branch 0 taken 9 times.

✓ Branch 1 taken 1915533 times.

1915542

if (*weight1 != *weight2) return 1;

2151

1915533

wt_ind1++;

2152

1915533

wt_ind2++;

2153

1915533

weight1 += UCA900_DISTANCE_BETWEEN_WEIGHTS;

2154

1915533

weight2 += UCA900_DISTANCE_BETWEEN_WEIGHTS;

2155

}

2156

/*

2157

If either character is out of weights but we have equality so far,

2158

check if the other character has any non-ignorable weights left.

2159

*/

2160

2/2

✓ Branch 0 taken 42 times.

✓ Branch 1 taken 1915693 times.

1915735

for (; wt_ind1 < length1; wt_ind1++) {

2161

2/2

✓ Branch 0 taken 6 times.

✓ Branch 1 taken 36 times.

42

if (*weight1) return 1;

2162

36

weight1 += UCA900_DISTANCE_BETWEEN_WEIGHTS;

2163

}

2164

2/2

✓ Branch 0 taken 305 times.

✓ Branch 1 taken 1915461 times.

1915766

for (; wt_ind2 < length2; wt_ind2++) {

2165

2/2

✓ Branch 0 taken 232 times.

✓ Branch 1 taken 73 times.

305

if (*weight2) return 1;

2166

73

weight2 += UCA900_DISTANCE_BETWEEN_WEIGHTS;

2167

}

2168

}

2169

1915364

return 0;

}

/*

This function compares if two code points are the same.

2174

The sign +1 or -1 does not matter. The only

2175

important thing is that the result is 0 or not 0.

2176

This fact allows us to use memcmp() safely, on both

2177

little-endian and big-endian machines.

2178

*/

2179

2180

182086189

static int my_uca_charcmp(const CHARSET_INFO *cs, my_wc_t wc1, my_wc_t wc2) {

2181

2/2

✓ Branch 0 taken 33936011 times.

✓ Branch 1 taken 148150178 times.

182086189

if (wc1 == wc2) return 0;

2182

2183

3/4

✓ Branch 0 taken 148150178 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 148148069 times.

✓ Branch 3 taken 2109 times.

148150178

if (cs->uca != nullptr && cs->uca->version == UCA_V900)

2184

148148069

return my_uca_charcmp_900(cs, wc1, wc2);

2185

2186

size_t length1, length2;

2187

2109

uint16 *weight1 = my_char_weight_addr(cs->uca, wc1); /* W3-TODO */

2188

2109

uint16 *weight2 = my_char_weight_addr(cs->uca, wc2);

2189

2190

/* Check if some of the code points does not have implicit weights */

2191

2/4

✓ Branch 0 taken 2109 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 2109 times.

2109

if (!weight1 || !weight2) return wc1 != wc2;

2192

2193

/* Quickly compare first weights */

2194

2/2

✓ Branch 0 taken 2073 times.

✓ Branch 1 taken 36 times.

2109

if (weight1[0] != weight2[0]) return 1;

2195

2196

/* Thoroughly compare all weights */

2197

36

length1 = cs->uca->lengths[wc1 >> MY_UCA_PSHIFT]; /* W3-TODO */

2198

36

length2 = cs->uca->lengths[wc2 >> MY_UCA_PSHIFT];

2199

2200

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 36 times.

36

if (length1 > length2)

2201

✗

return memcmp((const void *)weight1, (const void *)weight2, length2 * 2)

2202

✗

? 1

2203

✗

: weight1[length2];

2204

2205

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 36 times.

36

if (length1 < length2)

2206

✗

return memcmp((const void *)weight1, (const void *)weight2, length1 * 2)

2207

✗

? 1

2208

✗

: weight2[length1];

2209

2210

36

return memcmp((const void *)weight1, (const void *)weight2, length1 * 2);

2211

}

2212

2213

/*** Compare string against string with wildcard

2214

** 0 if matched

2215

** -1 if not matched with wildcard

2216

** 1 if matched with wildcard

2217

*/

2218

2219

131829251

static int my_wildcmp_uca_impl(const CHARSET_INFO *cs, const char *str,

2220

const char *str_end, const char *wildstr,

2221

const char *wildend, int escape, int w_one,

2222

int w_many, int recurse_level) {

2223

4/6

✓ Branch 0 taken 131828825 times.

✓ Branch 1 taken 426 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 131828825 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 131829251 times.

131829251

if (my_string_stack_guard && my_string_stack_guard(recurse_level)) return 1;

2224

2/2

✓ Branch 0 taken 131828585 times.

✓ Branch 1 taken 666 times.

131829251

while (wildstr != wildend) {

2225

131828585

int result = -1; /* Not found, using wildcards */

2226

131828585

auto mb_wc = cs->cset->mb_wc;

2227

2228

/*

2229

Compare the expression and pattern strings character-by-character until

2230

we find a '%' (w_many) in the pattern string. Once we do, we break out

2231

of the loop and try increasingly large widths for the '%' match,

2232

calling ourselves recursively until we find a match. (As an

2233

optimization, we test for the character immediately after '%' before we

2234

recurse.) This takes exponential time in the worst case.

2235

2236

Example: Say we are trying to match the pattern 'ab%cd' against the

2237

string 'ab..c.cd'. We first match the initial 'ab' against each other,

2238

and then see the '%' in the pattern. Since the first character after

2239

'%' is 'c', we skip to the first 'c' in the expression string, and try

2240

to match 'c.cd' against 'cd' by a recursive call. Since this failed, we

2241

scan for the next 'c', and try to match 'cd' against 'cd', which works.

*/

my_wc_t w_wc;

while (true) {

int mb_len;

1/2

✓ Branch 0 taken 168104851 times.

✗ Branch 1 not taken.

168104851

if ((mb_len = mb_wc(cs, &w_wc, (const uchar *)wildstr,

2247

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 168104851 times.

168104851

(const uchar *)wildend)) <= 0)

2248

130221051

return 1;

2249

2250

168104851

wildstr += mb_len;

2251

// If we found '%' (w_many), break out this loop.

2252

2/2

✓ Branch 0 taken 1607534 times.

✓ Branch 1 taken 166497317 times.

168104851

if (w_wc == (my_wc_t)w_many) {

2253

1607534

result = 1;

2254

1607534

break;

}

/*

If the character we just read was an escape character, skip it and

2259

read the next character instead. This character is used verbatim

2260

without checking if it is a wildcard (% or _). However, as a

2261

special exception, a lone escape character at the end of a string is

2262

treated as itself.

2263

*/

2264

166497317

bool escaped = false;

2265

4/4

✓ Branch 0 taken 7536 times.

✓ Branch 1 taken 166489781 times.

✓ Branch 2 taken 7522 times.

✓ Branch 3 taken 14 times.

166497317

if (w_wc == (my_wc_t)escape && wildstr < wildend) {

2266

1/2

✓ Branch 0 taken 7522 times.

✗ Branch 1 not taken.

7522

if ((mb_len = mb_wc(cs, &w_wc, (const uchar *)wildstr,

2267

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 7522 times.

7522

(const uchar *)wildend)) <= 0)

2268

✗

return 1;

2269

7522

wildstr += mb_len;

2270

7522

escaped = true;

}

my_wc_t s_wc;

1/2

✓ Branch 0 taken 166497317 times.

✗ Branch 1 not taken.

166497317

if ((mb_len = mb_wc(cs, &s_wc, (const uchar *)str,

2275

2/2

✓ Branch 0 taken 17541 times.

✓ Branch 1 taken 166479776 times.

166497317

(const uchar *)str_end)) <= 0)

2276

17541

return 1;

2277

166479776

str += mb_len;

2278

2279

// If we found '_' (w_one), skip one character in expression string.

2280

4/4

✓ Branch 0 taken 166472270 times.

✓ Branch 1 taken 7506 times.

✓ Branch 2 taken 2381496 times.

✓ Branch 3 taken 164090774 times.

166479776

if (!escaped && w_wc == (my_wc_t)w_one) {

2281

2381496

result = 1;

2282

} else {

2283

2/2

✓ Branch 0 taken 129958894 times.

✓ Branch 1 taken 34139386 times.

164098280

if (my_uca_charcmp(cs, s_wc, w_wc)) return 1;

2284

}

2285

2/2

✓ Branch 0 taken 244616 times.

✓ Branch 1 taken 36276266 times.

36520882

if (wildstr == wildend)

2286

244616

return (str != str_end); /* Match if both are at end */

2287

36276266

}

2288

2289

1/2

✓ Branch 0 taken 1607534 times.

✗ Branch 1 not taken.

1607534

if (w_wc == (my_wc_t)w_many) {

2290

// Remove any '%' and '_' following w_many in the pattern string.

2291

for (;;) {

2292

2/2

✓ Branch 0 taken 534622 times.

✓ Branch 1 taken 1086683 times.

1621305

if (wildstr == wildend) {

2293

/*

2294

The previous w_many (%) was the last character in the pattern

2295

string, so we have a match no matter what the rest of the

2296

expression string looks like (even empty).

2297

*/

2298

534622

return 0;

2299

}

2300

int mb_len_wild =

2301

1/2

✓ Branch 0 taken 1086683 times.

✗ Branch 1 not taken.

1086683

mb_wc(cs, &w_wc, (const uchar *)wildstr, (const uchar *)wildend);

2302

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1086683 times.

1086683

if (mb_len_wild <= 0) return 1;

2303

1086683

wildstr += mb_len_wild;

2304

2/2

✓ Branch 0 taken 2601 times.

✓ Branch 1 taken 1084082 times.

1086683

if (w_wc == (my_wc_t)w_many) continue;

2305

2306

2/2

✓ Branch 0 taken 11170 times.

✓ Branch 1 taken 1072912 times.

1084082

if (w_wc == (my_wc_t)w_one) {

2307

/*

2308

Skip one character in expression string because '_' needs to

match one.

*/

my_wc_t s_wc;

int mb_len =

1/2

✓ Branch 0 taken 11170 times.

✗ Branch 1 not taken.

11170

mb_wc(cs, &s_wc, (const uchar *)str, (const uchar *)str_end);

2314

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 11170 times.

11170

if (mb_len <= 0) return 1;

2315

11170

str += mb_len;

2316

11170

continue;

2317

11170

}

2318

1072912

break; /* Not a wild character */

2319

13771

}

2320

2321

// No character in the expression string to match w_wc.

2322

2/2

✓ Branch 0 taken 628 times.

✓ Branch 1 taken 1072284 times.

1072912

if (str == str_end) return -1;

2323

2324

// Skip the escape character ('\') in the pattern if needed.

2325

4/4

✓ Branch 0 taken 126 times.

✓ Branch 1 taken 1072158 times.

✓ Branch 2 taken 113 times.

✓ Branch 3 taken 13 times.

1072284

if (w_wc == (my_wc_t)escape && wildstr < wildend) {

2326

int mb_len =

2327

1/2

✓ Branch 0 taken 113 times.

✗ Branch 1 not taken.

113

mb_wc(cs, &w_wc, (const uchar *)wildstr, (const uchar *)wildend);

2328

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 113 times.

113

if (mb_len <= 0) return 1;

2329

113

wildstr += mb_len;

}

/*

w_wc is now the character following w_many (e.g., if the pattern is

"a%c", w_wc is 'c').

*/

while (true) {

/*

Skip until we find a character in the expression string that is

2339

equal to w_wc.

2340

*/

2341

2540992

int mb_len = 0;

2342

2/2

✓ Branch 0 taken 17987909 times.

✓ Branch 1 taken 828975 times.

18816884

while (str != str_end) {

2343

my_wc_t s_wc;

2344

1/2

✓ Branch 0 taken 17987909 times.

✗ Branch 1 not taken.

17987909

if ((mb_len = mb_wc(cs, &s_wc, (const uchar *)str,

2345

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 17987909 times.

17987909

(const uchar *)str_end)) <= 0)

2346

✗

return 1;

2347

2348

2/2

✓ Branch 0 taken 1712017 times.

✓ Branch 1 taken 16275892 times.

17987909

if (!my_uca_charcmp(cs, s_wc, w_wc)) break;

2349

16275892

str += mb_len;

2350

}

2351

// No character in the expression string is equal to w_wc.

2352

2/2

✓ Branch 0 taken 828975 times.

✓ Branch 1 taken 1712017 times.

2540992

if (str == str_end) return -1;

2353

1712017

str += mb_len;

2354

2355

/*

2356

The strings match up until the first character after w_many in the

2357

pattern string. For the rest part of pattern string and expression

2358

string, we recursively call to get wild compare result.

2359

Example, wildcmp(..., "abcdefg", "a%de%g", ...), we'll run again on

2360

wildcmp(..., "efg", "e%g", ...).

2361

*/

2362

1/2

✓ Branch 0 taken 1712017 times.

✗ Branch 1 not taken.

1712017

result = my_wildcmp_uca_impl(cs, str, str_end, wildstr, wildend, escape,

2363

w_one, w_many, recurse_level + 1);

2364

2365

2/2

✓ Branch 0 taken 243309 times.

✓ Branch 1 taken 1468708 times.

1712017

if (result <= 0) return result;

2366

1468708

}

2367

}

2368

}

2369

2/2

✓ Branch 0 taken 588 times.

✓ Branch 1 taken 78 times.

666

return (str != str_end ? 1 : 0);

2370

}

2371

2372

15962

static int my_strcasecmp_uca(const CHARSET_INFO *cs, const char *s,

2373

const char *t) {

2374

15962

const MY_UNICASE_INFO *uni_plane = cs->caseinfo;

2375

const MY_UNICASE_CHARACTER *page;

2376

4/4

✓ Branch 0 taken 34282 times.

✓ Branch 1 taken 3075 times.

✓ Branch 2 taken 33823 times.

✓ Branch 3 taken 459 times.

37357

while (s[0] && t[0]) {

2377

my_wc_t s_wc, t_wc;

2378

2379

1/2

✓ Branch 0 taken 33823 times.

✗ Branch 1 not taken.

33823

if (static_cast<uchar>(s[0]) < 128) {

2380

33823

s_wc = uni_plane->page[0][static_cast<uchar>(s[0])].tolower;

2381

33823

s++;

} else {

int res;

✗

res = cs->cset->mb_wc(cs, &s_wc, pointer_cast<const uchar *>(s),

2386

✗

pointer_cast<const uchar *>(s + 4));

2387

2388

0/2

✗ Branch 0 not taken.

✗ Branch 1 not taken.

12428

if (res <= 0) return strcmp(s, t);

2389

✗

s += res;

2390

✗

if (s_wc <= uni_plane->maxchar && (page = uni_plane->page[s_wc >> 8]))

2391

✗

s_wc = page[s_wc & 0xFF].tolower;

2392

}

2393

2394

/* Do the same for the second string */

2395

2396

1/2

✓ Branch 0 taken 33823 times.

✗ Branch 1 not taken.

33823

if (static_cast<uchar>(t[0]) < 128) {

2397

/* Convert single byte character into weight */

2398

33823

t_wc = uni_plane->page[0][static_cast<uchar>(t[0])].tolower;

2399

33823

t++;

2400

} else {

2401

✗

int res = cs->cset->mb_wc(cs, &t_wc, pointer_cast<const uchar *>(t),

2402

✗

pointer_cast<const uchar *>(t + 4));

2403

✗

if (res <= 0) return strcmp(s, t);

2404

✗

t += res;

2405

2406

✗

if (t_wc <= uni_plane->maxchar && (page = uni_plane->page[t_wc >> 8]))

2407

✗

t_wc = page[t_wc & 0xFF].tolower;

2408

}

2409

2410

/* Now we have two weights, let's compare them */

2411

2/2

✓ Branch 0 taken 12428 times.

✓ Branch 1 taken 21395 times.

33823

if (s_wc != t_wc) return static_cast<int>(s_wc) - static_cast<int>(t_wc);

2412

}

2413

3534

return static_cast<int>(static_cast<uchar>(s[0])) -

2414

3534

static_cast<int>(static_cast<uchar>(t[0]));

}

extern "C" {

130117234

static int my_wildcmp_uca(const CHARSET_INFO *cs, const char *str,

2419

const char *str_end, const char *wildstr,

2420

const char *wildend, int escape, int w_one,

2421

int w_many) {

2422

130117234

return my_wildcmp_uca_impl(cs, str, str_end, wildstr, wildend, escape, w_one,

2423

130117234

w_many, 1);

}

} // extern "C"

/*

Collation language is implemented according to

2429

subset of ICU Collation Customization (tailorings):

2430

http://icu.sourceforge.net/userguide/Collate_Customization.html

2431

2432

Collation language elements:

Delimiters:

space - skipped

<char> := A-Z | a-z | \uXXXX

2437

2438

Shift command:

2439

<shift> := & - reset at this letter.

2440

2441

Diff command:

2442

<d1> := < - Identifies a primary difference.

2443

<d2> := << - Identifies a secondary difference.

2444

<d3> := <<< - Idenfifies a tertiary difference.

Collation rules:

<ruleset> := <rule> { <ruleset> }

| <d2> <string>

| <d3> <string>

| <shift> <char>

<string> := <char> [ <string> ]

2456

2457

An example, Polish collation:

2458

2459

&A < \u0105 <<< \u0104

2460

&C < \u0107 <<< \u0106

2461

&E < \u0119 <<< \u0118

2462

&L < \u0142 <<< \u0141

2463

&N < \u0144 <<< \u0143

2464

&O < \u00F3 <<< \u00D3

2465

&S < \u015B <<< \u015A

2466

&Z < \u017A <<< \u017B

2467

*/

2468

2469

typedef enum my_coll_lexem_num_en {

2470

MY_COLL_LEXEM_EOF = 0,

2471

MY_COLL_LEXEM_SHIFT = 1,

2472

MY_COLL_LEXEM_RESET = 4,

2473

MY_COLL_LEXEM_CHAR = 5,

2474

MY_COLL_LEXEM_ERROR = 6,

2475

MY_COLL_LEXEM_OPTION = 7,

2476

MY_COLL_LEXEM_EXTEND = 8,

2477

MY_COLL_LEXEM_CONTEXT = 9

} my_coll_lexem_num;

/**

Convert collation customization lexem to string,

2482

for nice error reporting

2483

2484

@param term lexem code

@return lexem name

*/

1

static const char *my_coll_lexem_num_to_str(my_coll_lexem_num term) {

2490

1/9

✗ Branch 0 not taken.

✓ Branch 1 taken 1 times.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

1

switch (term) {

2491

✗

case MY_COLL_LEXEM_EOF:

2492

✗

return "EOF";

2493

1

case MY_COLL_LEXEM_SHIFT:

2494

1

return "Shift";

2495

✗

case MY_COLL_LEXEM_RESET:

2496

✗

return "&";

2497

✗

case MY_COLL_LEXEM_CHAR:

2498

✗

return "Character";

2499

✗

case MY_COLL_LEXEM_OPTION:

2500

✗

return "Bracket option";

2501

✗

case MY_COLL_LEXEM_EXTEND:

2502

✗

return "/";

2503

✗

case MY_COLL_LEXEM_CONTEXT:

2504

✗

return "|";

2505

✗

case MY_COLL_LEXEM_ERROR:

2506

✗

return "ERROR";

2507

}

2508

✗

return nullptr;

2509

}

2510

2511

struct MY_COLL_LEXEM {

2512

my_coll_lexem_num term;

const char *beg;

const char *end;

const char *prev;

int diff;

int code;

};

/*

Initialize collation rule lexical anilizer

SYNOPSIS

my_coll_lexem_init

lexem Lex analyzer to init

2526

str Const string to parse

2527

str_end End of the string

USAGE

RETURN VALUES

N/A

*/

29009

static void my_coll_lexem_init(MY_COLL_LEXEM *lexem, const char *str,

2535

const char *str_end) {

2536

29009

lexem->beg = str;

2537

29009

lexem->prev = str;

2538

29009

lexem->end = str_end;

2539

29009

lexem->diff = 0;

2540

29009

lexem->code = 0;

2541

29009

}

2542

2543

/**

2544

Compare lexem to string with length

2545

2546

@param lexem lexem

2547

@param pattern string

2548

@param patternlen string length

2549

2550

@retval 0 if lexem is equal to string, non-0 otherwise.

2551

*/

2552

2553

125963

static int lex_cmp(MY_COLL_LEXEM *lexem, const char *pattern,

2554

size_t patternlen) {

2555

125963

size_t lexemlen = lexem->beg - lexem->prev;

2556

2/2

✓ Branch 0 taken 62805 times.

✓ Branch 1 taken 63158 times.

125963

if (lexemlen < patternlen) return 1; /* Not a prefix */

2557

63158

return native_strncasecmp(lexem->prev, pattern, patternlen);

}

/*

Print collation customization expression parse error, with context.

2562

2563

SYNOPSIS

2564

my_coll_lexem_print_error

2565

lexem Lex analyzer to take context from

2566

errstr string to write error to

2567

errsize errstr size

2568

txt error message

2569

col_name collation name

USAGE

RETURN VALUES

N/A

*/

3

static void my_coll_lexem_print_error(MY_COLL_LEXEM *lexem, char *errstr,

2577

size_t errsize, const char *txt,

2578

const char *col_name) {

2579

char tail[30];

2580

3

size_t len = lexem->end - lexem->prev;

2581

1/2

✓ Branch 0 taken 3 times.

✗ Branch 1 not taken.

3

strmake(tail, lexem->prev, std::min(len, sizeof(tail) - 1));

2582

3

errstr[errsize - 1] = '\0';

2583

3

snprintf(errstr, errsize - 1, "%s at '%s' for COLLATION : %s",

2584

2/2

✓ Branch 0 taken 1 times.

✓ Branch 1 taken 2 times.

3

txt[0] ? txt : "Syntax error", tail, col_name);

2585

3

}

2586

2587

/*

2588

Convert a hex digit into its numeric value

SYNOPSIS

ch2x

ch hex digit to convert

USAGE

RETURN VALUES

an integer value in the range 0..15

-1 on error

*/

6647976

static int ch2x(int ch) {

2601

4/4

✓ Branch 0 taken 5881834 times.

✓ Branch 1 taken 766142 times.

✓ Branch 2 taken 3574586 times.

✓ Branch 3 taken 2307248 times.

6647976

if (ch >= '0' && ch <= '9') return ch - '0';

2602

2603

4/4

✓ Branch 0 taken 208941 times.

✓ Branch 1 taken 2864449 times.

✓ Branch 2 taken 4482 times.

✓ Branch 3 taken 204459 times.

3073390

if (ch >= 'a' && ch <= 'f') return 10 + ch - 'a';

2604

2605

4/4

✓ Branch 0 taken 1967559 times.

✓ Branch 1 taken 1101349 times.

✓ Branch 2 taken 1752102 times.

✓ Branch 3 taken 215457 times.

3068908

if (ch >= 'A' && ch <= 'F') return 10 + ch - 'A';

2606

2607

1316806

return -1;

}

/*

Collation language lexical parser:

2612

Scans the next lexem.

SYNOPSIS

my_coll_lexem_next

lexem Lex analyzer, previously initialized by

2617

my_coll_lexem_init.

2618

USAGE

2619

Call this function in a loop

2620

2621

RETURN VALUES

2622

Lexem number: eof, diff, shift, char or error.

2623

*/

2624

2625

3480019

static my_coll_lexem_num my_coll_lexem_next(MY_COLL_LEXEM *lexem) {

2626

const char *beg;

2627

my_coll_lexem_num rc;

2628

2629

2/2

✓ Branch 0 taken 5070598 times.

✓ Branch 1 taken 87022 times.

5157620

for (beg = lexem->beg; beg < lexem->end; beg++) {

2630

8/8

✓ Branch 0 taken 1677601 times.

✓ Branch 1 taken 25980 times.

✓ Branch 2 taken 293472 times.

✓ Branch 3 taken 267214 times.

✓ Branch 4 taken 56538 times.

✓ Branch 5 taken 204479 times.

✓ Branch 6 taken 795021 times.

✓ Branch 7 taken 1750293 times.

5070598

switch (*beg) {

2631

1677601

case ' ':

case '\t':

case '\r':

case '\n':

1677601

continue;

2636

2637

25980

case '[': /* Bracket expression, e.g. "[optimize [a-z]]" */

2638

{

2639

size_t nbrackets; /* Indicates nested recursion level */

2640

1/2

✓ Branch 0 taken 235013 times.

✗ Branch 1 not taken.

235013

for (beg++, nbrackets = 1; beg < lexem->end; beg++) {

2641

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 235013 times.

235013

if (*beg == '[') /* Enter nested bracket expression */

2642

✗

nbrackets++;

2643

2/2

✓ Branch 0 taken 25980 times.

✓ Branch 1 taken 209033 times.

235013

else if (*beg == ']') {

2644

1/2

✓ Branch 0 taken 25980 times.

✗ Branch 1 not taken.

25980

if (--nbrackets == 0) {

2645

25980

rc = MY_COLL_LEXEM_OPTION;

2646

25980

beg++;

2647

25980

goto ex;

}

}

}

✗

rc = MY_COLL_LEXEM_ERROR;

2652

✗

goto ex;

2653

}

2654

2655

293472

case '&':

2656

293472

beg++;

2657

293472

rc = MY_COLL_LEXEM_RESET;

2658

293472

goto ex;

2659

2660

267214

case '=':

2661

267214

beg++;

2662

267214

lexem->diff = 0;

2663

267214

rc = MY_COLL_LEXEM_SHIFT;

2664

267214

goto ex;

2665

2666

56538

case '/':

2667

56538

beg++;

2668

56538

rc = MY_COLL_LEXEM_EXTEND;

2669

56538

goto ex;

2670

2671

204479

case '|':

2672

204479

beg++;

2673

204479

rc = MY_COLL_LEXEM_CONTEXT;

2674

204479

goto ex;

2675

2676

795021

case '<': /* Shift: '<' or '<<' or '<<<' or '<<<<' */

2677

{

2678

/* Scan up to 3 additional '<' characters */

2679

795021

for (beg++, lexem->diff = 1;

2680

4/6

✓ Branch 0 taken 1683590 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 888569 times.

✓ Branch 3 taken 795021 times.

✓ Branch 4 taken 888569 times.

✗ Branch 5 not taken.

1683590

(beg < lexem->end) && (*beg == '<') && (lexem->diff <= 3);

2681

888569

beg++, lexem->diff++)

2682

;

2683

795021

rc = MY_COLL_LEXEM_SHIFT;

2684

795021

goto ex;

2685

}

2686

1750293

default:

2687

1750293

break;

2688

1677601

}

2689

2690

/* Escaped character, e.g. \u1234 */

2691

4/6

✓ Branch 0 taken 1329975 times.

✓ Branch 1 taken 420318 times.

✓ Branch 2 taken 1329975 times.

✗ Branch 3 not taken.

✓ Branch 4 taken 1329975 times.

✗ Branch 5 not taken.

1750293

if ((*beg == '\\') && (beg + 2 < lexem->end) && (beg[1] == 'u') &&

2692

2/2

✓ Branch 0 taken 1329973 times.

✓ Branch 1 taken 2 times.

1329975

my_isxdigit(&my_charset_utf8_general_ci, beg[2])) {

2693

int ch;

2694

2695

1329973

beg += 2;

2696

1329973

lexem->code = 0;

2697

6/6

✓ Branch 0 taken 6647976 times.

✓ Branch 1 taken 13167 times.

✓ Branch 2 taken 5331170 times.

✓ Branch 3 taken 1316806 times.

✓ Branch 4 taken 5331170 times.

✓ Branch 5 taken 1329973 times.

6661143

while ((beg < lexem->end) && ((ch = ch2x(beg[0])) >= 0)) {

2698

5331170

lexem->code = (lexem->code << 4) + ch;

2699

5331170

beg++;

2700

}

2701

1329973

rc = MY_COLL_LEXEM_CHAR;

2702

1329973

goto ex;

}

/*

Unescaped single byte character:

2707

allow printable ASCII range except SPACE and

2708

special characters parsed above []<&/|=

2709

*/

2710

3/4

✓ Branch 0 taken 336208 times.

✓ Branch 1 taken 84112 times.

✓ Branch 2 taken 336208 times.

✗ Branch 3 not taken.

420320

if (*beg >= 0x21 && *beg <= 0x7E) {

2711

336208

lexem->code = *beg++;

2712

336208

rc = MY_COLL_LEXEM_CHAR;

2713

336208

goto ex;

2714

}

2715

2716

1/2

✓ Branch 0 taken 84112 times.

✗ Branch 1 not taken.

84112

if (((uchar)*beg) > 0x7F) /* Unescaped multibyte character */

2717

{

2718

84112

CHARSET_INFO *cs = &my_charset_utf8_general_ci;

2719

my_wc_t wc;

2720

1/2

✓ Branch 0 taken 84112 times.

✗ Branch 1 not taken.

84112

int nbytes = cs->cset->mb_wc(cs, &wc, pointer_cast<const uchar *>(beg),

2721

84112

pointer_cast<const uchar *>(lexem->end));

2722

1/2

✓ Branch 0 taken 84112 times.

✗ Branch 1 not taken.

84112

if (nbytes > 0) {

2723

84112

rc = MY_COLL_LEXEM_CHAR;

2724

84112

beg += nbytes;

2725

84112

lexem->code = (int)wc;

2726

84112

goto ex;

}

}

✗

rc = MY_COLL_LEXEM_ERROR;

2731

✗

goto ex;

2732

}

2733

87022

rc = MY_COLL_LEXEM_EOF;

2734

2735

3480019

ex:

2736

3480019

lexem->prev = lexem->beg;

2737

3480019

lexem->beg = beg;

2738

3480019

lexem->term = rc;

2739

3480019

return rc;

}

/*

Collation rule item

*/

#define MY_UCA_MAX_EXPANSION 6 /* Maximum expansion length */

2747

2748

struct MY_COLL_RULE {

2749

my_wc_t base[MY_UCA_MAX_EXPANSION]; /* Base character */

2750

my_wc_t curr[MY_UCA_MAX_CONTRACTION]; /* Current character */

2751

int diff[4]; /* Primary, Secondary, Tertiary, Quaternary difference */

2752

size_t before_level; /* "reset before" indicator */

bool with_context;

};

/**

Return length of the "reset" string of a rule.

2758

2759

@param r Collation customization rule

2760

2761

@return Length of r->base

2762

*/

2763

2764

1250794

static inline size_t my_coll_rule_reset_length(MY_COLL_RULE *r) {

2765

1250794

return my_wstrnlen(r->base, MY_UCA_MAX_EXPANSION);

}

/**

Return length of the "shift" string of a rule.

2770

2771

@param r Collation customization rule

2772

2773

@return Length of r->base

2774

*/

2775

2776

1250794

static inline size_t my_coll_rule_shift_length(MY_COLL_RULE *r) {

2777

1250794

return my_wstrnlen(r->curr, MY_UCA_MAX_CONTRACTION);

}

/**

Append new character to the end of a 0-terminated wide string.

2782

2783

@param wc Wide string

2784

@param limit Maximum possible result length

2785

@param code Character to add

2786

2787

@return 1 if character was added, 0 if string was too long

2788

*/

2789

2790

1750409

static int my_coll_rule_expand(my_wc_t *wc, size_t limit, my_wc_t code) {

2791

size_t i;

2792

1/2

✓ Branch 0 taken 1957313 times.

✗ Branch 1 not taken.

1957313

for (i = 0; i < limit; i++) {

2793

2/2

✓ Branch 0 taken 1750409 times.

✓ Branch 1 taken 206904 times.

1957313

if (wc[i] == 0) {

2794

1750409

wc[i] = code;

2795

1750409

return 1;

2796

}

2797

}

2798

✗

return 0;

}

/**

Initialize collation customization rule

@param r Rule

*/

293472

static void my_coll_rule_reset(MY_COLL_RULE *r) { memset(r, 0, sizeof(*r)); }

/*

Shift methods:

Simple: "&B < C" : weight('C') = weight('B') + 1

2812

Expand: weight('C') = { weight('B'), weight(last_non_ignorable) + 1 }

2813

*/

2814

typedef enum {

2815

my_shift_method_simple = 0,

2816

my_shift_method_expand

2817

} my_coll_shift_method;

2818

2819

struct MY_COLL_RULES {

2820

MY_UCA_INFO *uca; /* Unicode weight data */

2821

size_t nrules; /* Number of rules in the rule array */

2822

size_t mrules; /* Number of allocated rules */

2823

MY_COLL_RULE *rule; /* Rule array */

2824

MY_CHARSET_LOADER *loader;

2825

my_coll_shift_method shift_after_method;

};

/**

Realloc rule array to a new size.

2830

Reallocate memory for 128 additional rules at once,

2831

to reduce the number of reallocs, which is important

2832

for long tailorings (e.g. for East Asian collations).

2833

2834

@param rules Rule container

2835

@param n new number of rules

2836

2837

@return 0 on success, -1 on error.

2838

*/

2839

2840

1250794

static int my_coll_rules_realloc(MY_COLL_RULES *rules, size_t n) {

2841

3/4

✓ Branch 0 taken 30883 times.

✓ Branch 1 taken 1219911 times.

✓ Branch 2 taken 1250794 times.

✗ Branch 3 not taken.

1281677

if (rules->nrules < rules->mrules ||

2842

1/2

✓ Branch 0 taken 30883 times.

✗ Branch 1 not taken.

30883

(rules->rule = static_cast<MY_COLL_RULE *>(rules->loader->mem_realloc(

2843

30883

rules->rule, sizeof(MY_COLL_RULE) * (rules->mrules = n + 128)))))

2844

1250794

return 0;

2845

✗

return -1;

}

/**

Append one new rule to a rule array

2850

2851

@param rules Rule container

2852

@param rule New rule to add

2853

2854

@return 0 on success, -1 on error.

2855

*/

2856

2857

1250794

static int my_coll_rules_add(MY_COLL_RULES *rules, MY_COLL_RULE *rule) {

2858

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1250794 times.

1250794

if (my_coll_rules_realloc(rules, rules->nrules + 1)) return -1;

2859

1250794

rules->rule[rules->nrules++] = rule[0];

2860

1250794

return 0;

}

/**

Apply difference at level

2865

2866

@param r Rule

2867

@param level Level (0,1,2,3,4)

2868

*/

2869

2870

1062235

static void my_coll_rule_shift_at_level(MY_COLL_RULE *r, int level) {

2871

5/6

✓ Branch 0 taken 37548 times.

✓ Branch 1 taken 341219 times.

✓ Branch 2 taken 93487 times.

✓ Branch 3 taken 322767 times.

✓ Branch 4 taken 267214 times.

✗ Branch 5 not taken.

1062235

switch (level) {

2872

37548

case 4: /* Quaternary difference */

2873

37548

r->diff[3]++;

2874

37548

break;

2875

341219

case 3: /* Tertiary difference */

2876

341219

r->diff[2]++;

2877

341219

r->diff[3] = 0;

2878

341219

break;

2879

93487

case 2: /* Secondary difference */

2880

93487

r->diff[1]++;

2881

93487

r->diff[2] = r->diff[3] = 0;

2882

93487

break;

2883

322767

case 1: /* Primary difference */

2884

322767

r->diff[0]++;

2885

322767

r->diff[1] = r->diff[2] = r->diff[3] = 0;

2886

322767

break;

2887

267214

case 0:

2888

/* Do nothing for '=': use the previous offsets for all levels */

2889

267214

break;

2890

✗

default:

2891

✗

assert(0);

2892

}

2893

1062235

}

2894

2895

struct MY_COLL_RULE_PARSER {

2896

MY_COLL_LEXEM tok[2]; /* Current token and next token for look-ahead */

2897

MY_COLL_RULE rule; /* Currently parsed rule */

2898

MY_COLL_RULES *rules; /* Rule list pointer */

2899

char errstr[128]; /* Error message */

};

/**

Current parser token

@param p Collation customization parser

2906

2907

@return Pointer to the current token

2908

*/

2909

2910

16265125

static MY_COLL_LEXEM *my_coll_parser_curr(MY_COLL_RULE_PARSER *p) {

2911

16265125

return &p->tok[0];

}

/**

Next parser token, to look ahead.

2916

2917

@param p Collation customization parser

2918

2919

@return Pointer to the next token

2920

*/

2921

2922

6902020

static MY_COLL_LEXEM *my_coll_parser_next(MY_COLL_RULE_PARSER *p) {

2923

6902020

return &p->tok[1];

}

/**

Scan one token from the input stream

2928

2929

@param p Collation customization parser

2930

2931

@return 1, for convenience, to use in logical expressions easier.

2932

*/

2933

3422001

static int my_coll_parser_scan(MY_COLL_RULE_PARSER *p) {

2934

3422001

my_coll_parser_curr(p)[0] = my_coll_parser_next(p)[0];

2935

3422001

my_coll_lexem_next(my_coll_parser_next(p));

2936

3422001

return 1;

}

/**

Initialize collation customization parser

2941

2942

@param p Collation customization parser

2943

@param rules Where to store rules

2944

@param str Beginning of a collation customization string

2945

@param str_end End of the collation customizations string

2946

*/

2947

2948

29009

static void my_coll_parser_init(MY_COLL_RULE_PARSER *p, MY_COLL_RULES *rules,

2949

const char *str, const char *str_end) {

2950

/*

2951

Initialize parser to the input buffer and scan two tokens,

2952

to make the current token and the next token known.

2953

*/

2954

29009

memset(p, 0, sizeof(*p));

2955

29009

p->rules = rules;

2956

29009

p->errstr[0] = '\0';

2957

29009

my_coll_lexem_init(my_coll_parser_curr(p), str, str_end);

2958

29009

my_coll_lexem_next(my_coll_parser_curr(p));

2959

29009

my_coll_parser_next(p)[0] = my_coll_parser_curr(p)[0];

2960

29009

my_coll_lexem_next(my_coll_parser_next(p));

2961

29009

}

2962

2963

/**

2964

Display error when an unexpected token found

2965

2966

@param p Collation customization parser

2967

@param term Which lexem was expected

2968

2969

@return 0, to use in "return" and boolean expressions.

2970

*/

2971

2972

1

static int my_coll_parser_expected_error(MY_COLL_RULE_PARSER *p,

2973

my_coll_lexem_num term) {

2974

1

snprintf(p->errstr, sizeof(p->errstr), "%s expected",

2975

my_coll_lexem_num_to_str(term));

2976

1

return 0;

}

/**

Display error when a too long character sequence is met

2981

2982

@param p Collation customization parser

2983

@param name Which kind of sequence: contraction, expansion, etc.

2984

2985

@return 0, to use in "return" and boolean expressions.

2986

*/

2987

2988

✗

static int my_coll_parser_too_long_error(MY_COLL_RULE_PARSER *p,

2989

const char *name) {

2990

✗

snprintf(p->errstr, sizeof(p->errstr), "%s is too long", name);

2991

✗

return 0;

}

/**

Scan the given lexem from input stream, or display "expected" error.

2996

2997

@param p Collation customization parser

2998

@param term Which lexem is expected.

2999

3000

@retval 0 if the required term was not found.

3001

@retval 1 if the required term was found.

3002

*/

3003

1939132

static int my_coll_parser_scan_term(MY_COLL_RULE_PARSER *p,

3004

my_coll_lexem_num term) {

3005

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1939132 times.

1939132

if (my_coll_parser_curr(p)->term != term)

3006

✗

return my_coll_parser_expected_error(p, term);

3007

1939132

return my_coll_parser_scan(p);

}

/*

In the following code we have a few functions to parse

3012

various collation customization non-terminal symbols.

3013

Unlike our usual coding convention, they return

3014

- 0 on "error" (when the rule was not scanned) and

3015

- 1 on "success"(when the rule was scanned).

3016

This is done intentionally to make body of the functions look easier

3017

and repeat the grammar of the rules in straightforward manner.

For example:

// <x> ::= <y> | <z>

int parse_x() { return parse_y() || parser_z(); }

3022

3023

// <x> ::= <y> <z>

3024

int parse_x() { return parse_y() && parser_z(); }

3025

3026

Using 1 on "not found" and 0 on "found" in the parser code would

3027

make the code more error prone and harder to read because

3028

of having to use inverse boolean logic.

*/

/**

Scan a collation setting in brakets, for example UCA version.

3033

3034

@param p Collation customization parser

3035

3036

@retval 0 if setting was scanned.

3037

@retval 1 if setting was not scanned.

3038

*/

3039

3040

11

static int my_coll_parser_scan_setting(MY_COLL_RULE_PARSER *p) {

3041

11

MY_COLL_RULES *rules = p->rules;

3042

11

MY_COLL_LEXEM *lexem = my_coll_parser_curr(p);

3043

3044

2/2

✓ Branch 0 taken 5 times.

✓ Branch 1 taken 6 times.

11

if (!lex_cmp(lexem, STRING_WITH_LEN("[version 4.0.0]"))) {

3045

5

rules->uca = &my_uca_v400;

3046

2/2

✓ Branch 0 taken 2 times.

✓ Branch 1 taken 4 times.

6

} else if (!lex_cmp(lexem, STRING_WITH_LEN("[version 5.2.0]"))) {

3047

2

rules->uca = &my_uca_v520;

3048

2/2

✓ Branch 0 taken 2 times.

✓ Branch 1 taken 2 times.

4

} else if (!lex_cmp(lexem, STRING_WITH_LEN("[shift-after-method expand]"))) {

3049

2

rules->shift_after_method = my_shift_method_expand;

3050

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 2 times.

2

} else if (!lex_cmp(lexem, STRING_WITH_LEN("[shift-after-method simple]"))) {

3051

✗

rules->shift_after_method = my_shift_method_simple;

3052

} else {

3053

2

return 0;

3054

}

3055

9

return my_coll_parser_scan(p);

}

/**

Scan multiple collation settings

3060

3061

@param p Collation customization parser

3062

3063

@retval 0 if no settings were scanned.

3064

@retval 1 if one or more settings were scanned.

3065

*/

3066

3067

29009

static int my_coll_parser_scan_settings(MY_COLL_RULE_PARSER *p) {

3068

/* Scan collation setting or special purpose command */

3069

2/2

✓ Branch 0 taken 11 times.

✓ Branch 1 taken 29007 times.

29018

while (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_OPTION) {

3070

2/2

✓ Branch 0 taken 2 times.

✓ Branch 1 taken 9 times.

11

if (!my_coll_parser_scan_setting(p)) return 0;

3071

}

3072

29007

return 1;

}

/**

Scan [before xxx] reset option

3077

3078

@param p Collation customization parser

3079

3080

@retval 0 if reset option was not scanned.

3081

@retval 1 if reset option was scanned.

3082

*/

3083

3084

25951

static int my_coll_parser_scan_reset_before(MY_COLL_RULE_PARSER *p) {

3085

25951

MY_COLL_LEXEM *lexem = my_coll_parser_curr(p);

3086

6/6

✓ Branch 0 taken 25907 times.

✓ Branch 1 taken 44 times.

✓ Branch 2 taken 6958 times.

✓ Branch 3 taken 18949 times.

✓ Branch 4 taken 7002 times.

✓ Branch 5 taken 18949 times.

51858

if (!lex_cmp(lexem, STRING_WITH_LEN("[before primary]")) ||

3087

25907

!lex_cmp(lexem, STRING_WITH_LEN("[before 1]"))) {

3088

7002

p->rule.before_level = 1;

3089

6/6

✓ Branch 0 taken 18945 times.

✓ Branch 1 taken 4 times.

✓ Branch 2 taken 1120 times.

✓ Branch 3 taken 17825 times.

✓ Branch 4 taken 1124 times.

✓ Branch 5 taken 17825 times.

37894

} else if (!lex_cmp(lexem, STRING_WITH_LEN("[before secondary]")) ||

3090

18945

!lex_cmp(lexem, STRING_WITH_LEN("[before 2]"))) {

3091

1124

p->rule.before_level = 2;

3092

6/6

✓ Branch 0 taken 17821 times.

✓ Branch 1 taken 4 times.

✓ Branch 2 taken 17765 times.

✓ Branch 3 taken 56 times.

✓ Branch 4 taken 17769 times.

✓ Branch 5 taken 56 times.

35646

} else if (!lex_cmp(lexem, STRING_WITH_LEN("[before tertiary]")) ||

3093

17821

!lex_cmp(lexem, STRING_WITH_LEN("[before 3]"))) {

3094

17769

p->rule.before_level = 3;

3095

5/6

✓ Branch 0 taken 52 times.

✓ Branch 1 taken 4 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 52 times.

✓ Branch 4 taken 4 times.

✓ Branch 5 taken 52 times.

108

} else if (!lex_cmp(lexem, STRING_WITH_LEN("[before quaternary]")) ||

3096

52

!lex_cmp(lexem, STRING_WITH_LEN("[before 4]"))) {

3097

4

p->rule.before_level = 4;

3098

} else {

3099

52

p->rule.before_level = 0;

3100

52

return 0; /* Don't scan the next character */

3101

}

3102

25899

return my_coll_parser_scan(p);

}

/**

Scan logical position and add to the wide string.

3107

3108

@param p Collation customization parser

3109

@param pwc Wide string to add code to

3110

@param limit The result string cannot be longer than 'limit' characters

3111

3112

@retval 0 if logical position was not scanned.

3113

@retval 1 if logical position was scanned.

3114

*/

3115

3116

70

static int my_coll_parser_scan_logical_position(MY_COLL_RULE_PARSER *p,

3117

my_wc_t *pwc, size_t limit) {

3118

70

MY_COLL_RULES *rules = p->rules;

3119

70

MY_COLL_LEXEM *lexem = my_coll_parser_curr(p);

3120

3121

2/2

✓ Branch 0 taken 8 times.

✓ Branch 1 taken 62 times.

70

if (!lex_cmp(lexem, STRING_WITH_LEN("[first non-ignorable]")))

3122

8

lexem->code = rules->uca->first_non_ignorable;

3123

2/2

✓ Branch 0 taken 12 times.

✓ Branch 1 taken 50 times.

62

else if (!lex_cmp(lexem, STRING_WITH_LEN("[last non-ignorable]")))

3124

12

lexem->code = rules->uca->last_non_ignorable;

3125

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 46 times.

50

else if (!lex_cmp(lexem, STRING_WITH_LEN("[first primary ignorable]")))

3126

4

lexem->code = rules->uca->first_primary_ignorable;

3127

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 42 times.

46

else if (!lex_cmp(lexem, STRING_WITH_LEN("[last primary ignorable]")))

3128

4

lexem->code = rules->uca->last_primary_ignorable;

3129

2/2

✓ Branch 0 taken 6 times.

✓ Branch 1 taken 36 times.

42

else if (!lex_cmp(lexem, STRING_WITH_LEN("[first secondary ignorable]")))

3130

6

lexem->code = rules->uca->first_secondary_ignorable;

3131

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 32 times.

36

else if (!lex_cmp(lexem, STRING_WITH_LEN("[last secondary ignorable]")))

3132

4

lexem->code = rules->uca->last_secondary_ignorable;

3133

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 28 times.

32

else if (!lex_cmp(lexem, STRING_WITH_LEN("[first tertiary ignorable]")))

3134

4

lexem->code = rules->uca->first_tertiary_ignorable;

3135

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 24 times.

28

else if (!lex_cmp(lexem, STRING_WITH_LEN("[last tertiary ignorable]")))

3136

4

lexem->code = rules->uca->last_tertiary_ignorable;

3137

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 20 times.

24

else if (!lex_cmp(lexem, STRING_WITH_LEN("[first trailing]")))

3138

4

lexem->code = rules->uca->first_trailing;

3139

2/2

✓ Branch 0 taken 4 times.

✓ Branch 1 taken 16 times.

20

else if (!lex_cmp(lexem, STRING_WITH_LEN("[last trailing]")))

3140

4

lexem->code = rules->uca->last_trailing;

3141

2/2

✓ Branch 0 taken 8 times.

✓ Branch 1 taken 8 times.

16

else if (!lex_cmp(lexem, STRING_WITH_LEN("[first variable]")))

3142

8

lexem->code = rules->uca->first_variable;

3143

1/2

✓ Branch 0 taken 8 times.

✗ Branch 1 not taken.

8

else if (!lex_cmp(lexem, STRING_WITH_LEN("[last variable]")))

3144

8

lexem->code = rules->uca->last_variable;

3145

else

3146

✗

return 0; /* Don't scan the next token */

3147

3148

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 70 times.

70

if (!my_coll_rule_expand(pwc, limit, lexem->code)) {

3149

/*

3150

Logical position can not be in a contraction,

3151

so the above call should never fail.

3152

Let's assert in debug version and print

3153

a nice error message in production version.

3154

*/

3155

✗

assert(0);

3156

return my_coll_parser_too_long_error(p, "Logical position");

3157

}

3158

70

return my_coll_parser_scan(p);

}

/**

Scan character list

@<character list@> ::= CHAR [ CHAR... ]

3165

3166

@param p Collation customization parser

3167

@param pwc Character string to add code to

3168

@param limit The result string cannot be longer than 'limit' characters

3169

@param name E.g. "contraction", "expansion"

3170

3171

@retval 0 if character sequence was not scanned.

3172

@retval 1 if character sequence was scanned.

3173

*/

3174

3175

1616654

static int my_coll_parser_scan_character_list(MY_COLL_RULE_PARSER *p,

3176

my_wc_t *pwc, size_t limit,

3177

const char *name) {

3178

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1616654 times.

1616654

if (my_coll_parser_curr(p)->term != MY_COLL_LEXEM_CHAR)

3179

✗

return my_coll_parser_expected_error(p, MY_COLL_LEXEM_CHAR);

3180

3181

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1616654 times.

1616654

if (!my_coll_rule_expand(pwc, limit, my_coll_parser_curr(p)->code))

3182

✗

return my_coll_parser_too_long_error(p, name);

3183

3184

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1616654 times.

1616654

if (!my_coll_parser_scan_term(p, MY_COLL_LEXEM_CHAR)) return 0;

3185

3186

2/2

✓ Branch 0 taken 133639 times.

✓ Branch 1 taken 1616654 times.

1750293

while (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_CHAR) {

3187

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 133639 times.

133639

if (!my_coll_rule_expand(pwc, limit, my_coll_parser_curr(p)->code))

3188

✗

return my_coll_parser_too_long_error(p, name);

3189

133639

my_coll_parser_scan(p);

3190

}

3191

1616654

return 1;

}

/**

Scan reset sequence

@<reset sequence@> ::=

3198

[ @<reset before option@> ] @<character list@>

3199

| [ @<reset before option@> ] @<logical reset position@>

3200

3201

@param p Collation customization parser

3202

3203

@retval 0 if reset sequence was not scanned.

3204

@retval 1 if reset sequence was scanned.

3205

*/

3206

3207

293472

static int my_coll_parser_scan_reset_sequence(MY_COLL_RULE_PARSER *p) {

3208

293472

my_coll_rule_reset(&p->rule);

3209

3210

/* Scan "[before x]" option, if exists */

3211

2/2

✓ Branch 0 taken 25951 times.

✓ Branch 1 taken 267521 times.

293472

if (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_OPTION)

3212

25951

my_coll_parser_scan_reset_before(p);

3213

3214

/* Try logical reset position */

3215

2/2

✓ Branch 0 taken 70 times.

✓ Branch 1 taken 293402 times.

293472

if (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_OPTION) {

3216

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 70 times.

70

if (!my_coll_parser_scan_logical_position(p, p->rule.base, 1)) return 0;

3217

} else {

3218

/* Scan single reset character or expansion */

3219

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 293402 times.

293402

if (!my_coll_parser_scan_character_list(p, p->rule.base,

3220

MY_UCA_MAX_EXPANSION, "Expansion"))

3221

✗

return 0;

3222

}

3223

3224

2/2

✓ Branch 0 taken 293468 times.

✓ Branch 1 taken 4 times.

293472

if ((p->rules->shift_after_method == my_shift_method_expand ||

3225

2/2

✓ Branch 0 taken 7000 times.

✓ Branch 1 taken 286468 times.

293468

p->rule.before_level == 1) &&

3226

2/2

✓ Branch 0 taken 46 times.

✓ Branch 1 taken 6958 times.

7004

p->rules->uca->version < UCA_V900) /* Apply "before primary" option */

3227

{

3228

/*

3229

Suppose we have this rule: &B[before primary] < C

3230

i.e. we need to put C before B, but after A, so

3231

the result order is: A < C < B.

3232

3233

Let primary weight of B be [BBBB].

3234

3235

We cannot just use [BBBB-1] as weight for C:

3236

DUCET does not have enough unused weights between any two characters,

3237

so using [BBBB-1] will likely make C equal to the previous character,

3238

which is A, so we'll get this order instead of the desired: A = C < B.

3239

3240

To guarantee that that C is sorted after A, we'll use expansion

3241

with a kind of "biggest possible character".

3242

As "biggest possible character" we'll use "last_non_ignorable":

3243

3244

We'll compose weight for C as: [BBBB-1][MMMM+1]

3245

where [MMMM] is weight for "last_non_ignorable".

3246

3247

We also do the same trick for "reset after" if the collation

3248

option says so. E.g. for the rules "&B < C", weight for

3249

C will be calculated as: [BBBB][MMMM+1]

3250

3251

At this point we only need to store codepoints

3252

'B' and 'last_non_ignorable'. Actual weights for 'C'

3253

will be calculated according to the above formula later,

3254

in create_tailoring().

3255

*/

3256

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 46 times.

46

if (!my_coll_rule_expand(p->rule.base, MY_UCA_MAX_EXPANSION,

3257

46

p->rules->uca->last_non_ignorable))

3258

✗

return my_coll_parser_too_long_error(p, "Expansion");

3259

}

3260

293472

return 1;

}

/**

Scan shift sequence

@<shift sequence@> ::=

3267

@<character list@> [ / @<character list@> ]

3268

| @<character list@> [ | @<character list@> ]

3269

3270

@param p Collation customization parser

3271

3272

@retval 0 if shift sequence was not scanned.

3273

@retval 1 if shift sequence was scanned.

3274

*/

3275

3276

1062235

static int my_coll_parser_scan_shift_sequence(MY_COLL_RULE_PARSER *p) {

3277

MY_COLL_RULE before_extend;

3278

3279

1062235

memset(&p->rule.curr, 0, sizeof(p->rule.curr));

3280

3281

/* Scan single shift character or contraction */

3282

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1062235 times.

1062235

if (!my_coll_parser_scan_character_list(

3283

1/2

✓ Branch 0 taken 1062235 times.

✗ Branch 1 not taken.

1062235

p, p->rule.curr, MY_UCA_MAX_CONTRACTION, "Contraction"))

3284

✗

return 0;

3285

3286

1062235

before_extend = p->rule; /* Remember the part before "/" */

3287

3288

/* Append the part after "/" as expansion */

3289

2/2

✓ Branch 0 taken 11641 times.

✓ Branch 1 taken 1050594 times.

1062235

if (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_EXTEND) {

3290

1/2

✓ Branch 0 taken 11641 times.

✗ Branch 1 not taken.

11641

my_coll_parser_scan(p);

3291

2/4

✓ Branch 0 taken 11641 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 11641 times.

11641

if (!my_coll_parser_scan_character_list(p, p->rule.base,

3292

MY_UCA_MAX_EXPANSION, "Expansion"))

3293

✗

return 0;

3294

2/2

✓ Branch 0 taken 204479 times.

✓ Branch 1 taken 846115 times.

1050594

} else if (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_CONTEXT) {

3295

/*

3296

We support 2-character long context sequences only:

3297

one character is the previous context, plus the current character.

3298

It's OK as Unicode's CLDR does not have longer examples.

3299

*/

3300

1/2

✓ Branch 0 taken 204479 times.

✗ Branch 1 not taken.

204479

my_coll_parser_scan(p);

3301

204479

p->rule.with_context = true;

3302

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 204479 times.

204479

if (!my_coll_parser_scan_character_list(

3303

1/2

✓ Branch 0 taken 204479 times.

✗ Branch 1 not taken.

204479

p, p->rule.curr + 1, MY_UCA_MAX_EXPANSION - 1, "context"))

3304

✗

return 0;

3305

/*

3306

It might be CONTEXT followed by EXPANSION. For example, Japanese

3307

collation has one rule defined as:

3308

"&[before 3]へ<<<へ|ゝ=べ|ゝ=へ|ゞ/\u3099"

3309

The part of "へ|ゞ/\u3099" is CONTEXT ('|') followed by EXPANSION ('/').

3310

*/

3311

2/2

✓ Branch 0 taken 44897 times.

✓ Branch 1 taken 159582 times.

204479

if (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_EXTEND) {

3312

1/2

✓ Branch 0 taken 44897 times.

✗ Branch 1 not taken.

44897

my_coll_parser_scan(p);

3313

44897

size_t len = my_wstrnlen(p->rule.base, MY_UCA_MAX_EXPANSION);

3314

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 44897 times.

44897

if (!my_coll_parser_scan_character_list(

3315

1/2

✓ Branch 0 taken 44897 times.

✗ Branch 1 not taken.

44897

p, p->rule.base + len, MY_UCA_MAX_EXPANSION - len, "Expansion"))

3316

✗

return 0;

}

}

/* Add rule to the rule list */

3321

2/4

✓ Branch 0 taken 1062235 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 1062235 times.

1062235

if (my_coll_rules_add(p->rules, &p->rule)) return 0;

3322

3323

1062235

p->rule = before_extend; /* Restore to the state before "/" */

3324

3325

1062235

return 1;

}

/**

Scan shift operator

@<shift@> ::= < | << | <<< | <<<< | =

3332

3333

@param p Collation customization parser

3334

3335

@retval 0 if shift operator was not scanned.

3336

@retval 1 if shift operator was scanned.

3337

*/

3338

1355707

static int my_coll_parser_scan_shift(MY_COLL_RULE_PARSER *p) {

3339

2/2

✓ Branch 0 taken 1062235 times.

✓ Branch 1 taken 293472 times.

1355707

if (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_SHIFT) {

3340

1062235

my_coll_rule_shift_at_level(&p->rule, my_coll_parser_curr(p)->diff);

3341

1062235

return my_coll_parser_scan(p);

3342

}

3343

293472

return 0;

}

/**

Scan one rule: reset followed by a number of shifts

@<rule@> ::=

& @<reset sequence@>

@<shift@> @<shift sequence@>

3352

[ { @<shift@> @<shift sequence@> }... ]

3353

3354

@param p Collation customization parser

3355

3356

@retval 0 if rule was not scanned.

3357

@retval 1 if rule was scanned.

3358

*/

3359

293472

static int my_coll_parser_scan_rule(MY_COLL_RULE_PARSER *p) {

3360

2/4

✓ Branch 0 taken 293472 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 293472 times.

586944

if (!my_coll_parser_scan_term(p, MY_COLL_LEXEM_RESET) ||

3361

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 293472 times.

293472

!my_coll_parser_scan_reset_sequence(p))

3362

✗

return 0;

3363

3364

/* Scan the first required shift command */

3365

2/2

✓ Branch 0 taken 1 times.

✓ Branch 1 taken 293471 times.

293472

if (!my_coll_parser_scan_shift(p))

3366

1

return my_coll_parser_expected_error(p, MY_COLL_LEXEM_SHIFT);

3367

3368

/* Scan the first shift sequence */

3369

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 293471 times.

293471

if (!my_coll_parser_scan_shift_sequence(p)) return 0;

3370

3371

/* Scan subsequent shift rules */

3372

2/2

✓ Branch 0 taken 768764 times.

✓ Branch 1 taken 293471 times.

1062235

while (my_coll_parser_scan_shift(p)) {

3373

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 768764 times.

768764

if (!my_coll_parser_scan_shift_sequence(p)) return 0;

3374

}

3375

293471

return 1;

}

/**

Scan collation customization: settings followed by rules

3380

3381

@<collation customization@> ::=

[ @<setting@> ... ]

[ @<rule@>... ]

@param p Collation customization parser

3386

3387

@retval 0 if collation customization expression was not scanned.

3388

@retval 1 if collation customization expression was scanned.

3389

*/

3390

3391

29009

static int my_coll_parser_exec(MY_COLL_RULE_PARSER *p) {

3392

2/2

✓ Branch 0 taken 2 times.

✓ Branch 1 taken 29007 times.

29009

if (!my_coll_parser_scan_settings(p)) return 0;

3393

3394

2/2

✓ Branch 0 taken 293472 times.

✓ Branch 1 taken 29006 times.

322478

while (my_coll_parser_curr(p)->term == MY_COLL_LEXEM_RESET) {

3395

2/2

✓ Branch 0 taken 1 times.

✓ Branch 1 taken 293471 times.

293472

if (!my_coll_parser_scan_rule(p)) return 0;

3396

}

3397

/* Make sure no unparsed input data left */

3398

29006

return my_coll_parser_scan_term(p, MY_COLL_LEXEM_EOF);

}

/*

Collation language syntax parser.

3403

Uses lexical parser.

3404

3405

@param rules Collation rule list to load to.

3406

@param str A string with collation customization.

3407

@param str_end End of the string.

3408

@param col_name Collation name

@retval 0 on success

@retval 1 on error

*/

29009

static int my_coll_rule_parse(MY_COLL_RULES *rules, const char *str,

3415

const char *str_end, const char *col_name) {

3416

MY_COLL_RULE_PARSER p;

3417

3418

1/2

✓ Branch 0 taken 29009 times.

✗ Branch 1 not taken.

29009

my_coll_parser_init(&p, rules, str, str_end);

3419

3420

3/4

✓ Branch 0 taken 29009 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 3 times.

✓ Branch 3 taken 29006 times.

29009

if (!my_coll_parser_exec(&p)) {

3421

3

rules->loader->errcode = EE_COLLATION_PARSER_ERROR;

3422

1/2

✓ Branch 0 taken 3 times.

✗ Branch 1 not taken.

3

my_coll_lexem_print_error(my_coll_parser_curr(&p), rules->loader->errarg,

3423

sizeof(rules->loader->errarg) - 1, p.errstr,

3424

col_name);

3425

3

return 1;

3426

}

3427

29006

return 0;

3428

}

3429

3430

7536

static void spread_case_mask(uint16 *to, size_t to_stride,

3431

size_t tailored_ce_cnt, uint16 case_mask) {

3432

2/2

✓ Branch 0 taken 18526 times.

✓ Branch 1 taken 7536 times.

26062

for (size_t i = 0; i < tailored_ce_cnt; ++i) {

3433

18526

uint16 *case_weight = &to[(i * MY_UCA_900_CE_SIZE + 2) * to_stride];

3434

2/2

✓ Branch 0 taken 8164 times.

✓ Branch 1 taken 10362 times.

18526

if (*case_weight > CASE_FIRST_UPPER_MASK)

3435

8164

case_mask = *case_weight & 0xFF00;

3436

2/2

✓ Branch 0 taken 6751 times.

✓ Branch 1 taken 3611 times.

10362

else if (*case_weight)

3437

6751

*case_weight |= case_mask;

3438

}

3439

7536

}

3440

3441

/*

3442

If the collation is marked as [caseFirst upper], move all of the weights

3443

around to accommodate that. Only tailored weights are changed; for non-tailored

3444

weights, we do it on-the-fly in uca_scanner_900::apply_case_first().

3445

3446

[caseFirst upper] is a directive that says that case should override all

3447

other tertiary case concerns (in a sense, a “level 2.5”), and furthermore,

3448

that uppercase should come before lowercase. (Normally lowercase sorts

3449

before uppercase.) It is currently only used in the Danish collation.

3450

3451

This is done by looking at the tertiary weight, inferring the case from it,

3452

and then using the upper bits (which are normally unused) to signal the case.

3453

The algorithm is detailed in Unicode TR35, section 3.14, although we don't

3454

seem to follow it exactly.

3455

*/

3456

1250794

static void change_weight_if_case_first(CHARSET_INFO *cs,

3457

const MY_UCA_INFO *dst, MY_COLL_RULE *r,

3458

uint16 *to, size_t to_stride,

3459

size_t curr_len,

3460

size_t tailored_ce_cnt) {

3461

/* We only need to implement [caseFirst upper] right now. */

3462

4/4

✓ Branch 0 taken 518711 times.

✓ Branch 1 taken 732083 times.

✓ Branch 2 taken 7536 times.

✓ Branch 3 taken 511175 times.

1250794

if (!(cs->coll_param && cs->coll_param->case_first == CASE_FIRST_UPPER &&

3463

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 7536 times.

7536

cs->levels_for_compare == 3))

3464

1243258

return;

3465

3466

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 7536 times.

7536

assert(cs->uca->version == UCA_V900);

3467

3468

// How many CEs this character has with non-ignorable primary weight.

3469

7536

int tailored_pri_cnt = 0;

3470

7536

int origin_pri_cnt = 0;

3471

2/2

✓ Branch 0 taken 12089 times.

✓ Branch 1 taken 4239 times.

16328

for (size_t i = 0; i < tailored_ce_cnt; ++i) {

3472

/*

3473

If rule A has already applied a case weight change, and we have rule B

3474

which is inherited from A, apply the same case weight change on the rest

3475

of rule B and return.

3476

*/

3477

2/2

✓ Branch 0 taken 3297 times.

✓ Branch 1 taken 8792 times.

12089

if (to[(i * MY_UCA_900_CE_SIZE + 2) * to_stride] > CASE_FIRST_UPPER_MASK) {

3478

3297

spread_case_mask(to, to_stride, tailored_ce_cnt, /*case_mask=*/0);

3479

3297

return;

3480

}

3481

2/2

✓ Branch 0 taken 7222 times.

✓ Branch 1 taken 1570 times.

8792

if (to[i * MY_UCA_900_CE_SIZE * to_stride]) tailored_pri_cnt++;

3482

}

3483

3/4

✓ Branch 0 taken 1570 times.

✓ Branch 1 taken 2669 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 1570 times.

4239

if (r->before_level == 1 || r->diff[0]) tailored_pri_cnt--;

3484

3485

// Use the DUCET weight to detect the character's case.

3486

4239

MY_UCA_INFO *src = &my_uca_v900;

3487

4239

int changed_ce = 0;

3488

3489

4239

my_wc_t *curr = r->curr;

3490

2/2

✓ Branch 0 taken 4710 times.

✓ Branch 1 taken 4239 times.

8949

for (size_t i = 0; i < curr_len; ++i) {

3491

4710

const uint16 *from = my_char_weight_addr_900(src, *curr);

3492

4710

uint page = *curr >> 8;

3493

4710

uint code = *curr & 0xFF;

3494

4710

curr++;

3495

4710

int ce_cnt =

3496

1/2

✓ Branch 0 taken 4710 times.

✗ Branch 1 not taken.

4710

src->weights[page] ? UCA900_NUM_OF_CE(src->weights[page], code) : 0;

3497

2/2

✓ Branch 0 taken 8792 times.

✓ Branch 1 taken 4710 times.

13502

for (int i_ce = 0; i_ce < ce_cnt; ++i_ce) {

3498

2/2

✓ Branch 0 taken 5338 times.

✓ Branch 1 taken 3454 times.

8792

if (from[i_ce * UCA900_DISTANCE_BETWEEN_WEIGHTS]) origin_pri_cnt++;

3499

}

3500

}

3501

4239

int case_to_copy = 0;

3502

2/2

✓ Branch 0 taken 3140 times.

✓ Branch 1 taken 1099 times.

4239

if (origin_pri_cnt <= tailored_pri_cnt)

3503

3140

case_to_copy = origin_pri_cnt;

3504

else

3505

1099

case_to_copy = tailored_pri_cnt - 1;

3506

4239

int upper_cnt = 0;

3507

4239

int lower_cnt = 0;

3508

4239

curr = r->curr;

3509

4239

uint16 case_mask = 0;

3510

2/2

✓ Branch 0 taken 4710 times.

✓ Branch 1 taken 4239 times.

8949

for (size_t curr_ind = 0; curr_ind < curr_len; ++curr_ind) {

3511

4710

const uint16 *from = my_char_weight_addr_900(src, *curr);

3512

4710

uint page = *curr >> 8;

3513

4710

uint code = *curr & 0xFF;

3514

4710

curr++;

3515

4710

int ce_cnt =

3516

1/2

✓ Branch 0 taken 4710 times.

✗ Branch 1 not taken.

4710

src->weights[page] ? UCA900_NUM_OF_CE(src->weights[page], code) : 0;

3517

4710

changed_ce = 0;

3518

2/2

✓ Branch 0 taken 8792 times.

✓ Branch 1 taken 4710 times.

13502

for (int i_ce = 0; i_ce < ce_cnt; ++i_ce) {

3519

8792

uint16 primary_weight = from[i_ce * UCA900_DISTANCE_BETWEEN_WEIGHTS];

3520

2/2

✓ Branch 0 taken 5338 times.

✓ Branch 1 taken 3454 times.

8792

if (primary_weight) {

3521

5338

uint16 case_weight = from[i_ce * UCA900_DISTANCE_BETWEEN_WEIGHTS +

3522

5338

2 * UCA900_DISTANCE_BETWEEN_LEVELS];

3523

5338

uint16 *ce_to = nullptr;

3524

2/2

✓ Branch 0 taken 2669 times.

✓ Branch 1 taken 2669 times.

5338

if (is_tertiary_weight_upper_case(case_weight)) {

3525

2/2

✓ Branch 0 taken 1099 times.

✓ Branch 1 taken 1570 times.

2669

if (!case_to_copy)

3526

1099

upper_cnt++;

3527

else

3528

1570

case_mask = CASE_FIRST_UPPER_MASK;

3529

} else {

3530

2/2

✓ Branch 0 taken 1099 times.

✓ Branch 1 taken 1570 times.

2669

if (!case_to_copy)

3531

1099

lower_cnt++;

3532

else

3533

1570

case_mask = CASE_FIRST_LOWER_MASK;

3534

}

3535

2/2

✓ Branch 0 taken 3140 times.

✓ Branch 1 taken 2198 times.

5338

if (case_to_copy) {

3536

do {

3537

3140

ce_to = to + changed_ce * MY_UCA_900_CE_SIZE * to_stride;

3538

3140

changed_ce++;

3539

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 3140 times.

3140

} while (*ce_to == 0);

3540

3140

ce_to[2 * to_stride] |= case_mask;

3541

3140

case_to_copy--;

}

}

}

}

2/2

✓ Branch 0 taken 3140 times.

✓ Branch 1 taken 1099 times.

4239

if (origin_pri_cnt <= tailored_pri_cnt) {

3547

2/2

✓ Branch 0 taken 314 times.

✓ Branch 1 taken 3140 times.

3454

for (int i = origin_pri_cnt; i < tailored_pri_cnt; ++i) {

3548

314

const int offset = changed_ce * MY_UCA_900_CE_SIZE * to_stride;

3549

2/4

✓ Branch 0 taken 314 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 314 times.

✗ Branch 3 not taken.

314

if (to[offset] && to[offset] < dst->extra_ce_pri_base)

3550

314

to[offset + 2 * to_stride] = 0;

3551

}

3552

} else {

3553

4/4

✓ Branch 0 taken 628 times.

✓ Branch 1 taken 471 times.

✓ Branch 2 taken 157 times.

✓ Branch 3 taken 471 times.

1099

if (upper_cnt && lower_cnt)

3554

157

case_mask = CASE_FIRST_MIXED_MASK;

3555

3/4

✓ Branch 0 taken 471 times.

✓ Branch 1 taken 471 times.

✓ Branch 2 taken 471 times.

✗ Branch 3 not taken.

942

else if (upper_cnt && !lower_cnt)

3556

471

case_mask = CASE_FIRST_UPPER_MASK;

3557

else

3558

471

case_mask = CASE_FIRST_LOWER_MASK;

3559

1099

bool skipped_extra_ce = false;

3560

2/2

✓ Branch 0 taken 2198 times.

✓ Branch 1 taken 1099 times.

3297

for (int i = tailored_ce_cnt - 1; i >= 0; --i) {

3561

2198

int offset = i * MY_UCA_900_CE_SIZE * to_stride;

3562

3/4

✓ Branch 0 taken 2198 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 1099 times.

✓ Branch 3 taken 1099 times.

2198

if (to[offset] && to[offset] < dst->extra_ce_pri_base) {

3563

2/6

✗ Branch 0 not taken.

✓ Branch 1 taken 1099 times.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✓ Branch 4 taken 1099 times.

✗ Branch 5 not taken.

1099

if ((r->before_level == 1 || r->diff[0]) && !skipped_extra_ce) {

3564

1099

skipped_extra_ce = true;

3565

1099

continue;

3566

}

3567

✗

to[(i * MY_UCA_900_CE_SIZE + 2) * to_stride] |= case_mask;

3568

✗

break;

}

}

}

4239

spread_case_mask(to, to_stride, tailored_ce_cnt, case_mask);

3573

}

3574

3575

734767

static size_t my_char_weight_put_900(MY_UCA_INFO *dst, uint16 *to,

3576

size_t to_stride, size_t to_length,

3577

uint16 *to_num_ce,

3578

const MY_COLL_RULE *rule,

3579

size_t base_len) {

3580

size_t count;

3581

734767

int total_ce_cnt = 0;

3582

3583

734767

const my_wc_t *base = rule->base;

3584

2/2

✓ Branch 0 taken 1034642 times.

✓ Branch 1 taken 734767 times.

1769409

for (count = 0; base_len;) {

3585

1034642

const uint16 *from = nullptr;

3586

1034642

size_t from_stride = 0;

3587

1034642

int ce_cnt = 0;

3588

3589

2/2

✓ Branch 0 taken 366380 times.

✓ Branch 1 taken 1015819 times.

1382199

for (size_t chlen = base_len; chlen > 1; chlen--) {

3590

3/4

✓ Branch 0 taken 366380 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 18823 times.

✓ Branch 3 taken 347557 times.

366380

if ((from = my_uca_contraction_weight(dst->contraction_nodes, base,

3591

chlen))) {

3592

18823

from_stride = 1;

3593

18823

base += chlen;

3594

18823

base_len -= chlen;

3595

18823

ce_cnt = *(from + MY_UCA_MAX_WEIGHT_SIZE - 1);

3596

18823

break;

}

}

2/2

✓ Branch 0 taken 1015819 times.

✓ Branch 1 taken 18823 times.

1034642

if (!from) {

3601

1015819

uint page = *base >> 8;

3602

1015819

uint code = *base & 0xFF;

3603

1015819

base++;

3604

1015819

base_len--;

3605

1/2

✓ Branch 0 taken 1015819 times.

✗ Branch 1 not taken.

1015819

if (dst->weights[page]) {

3606

1015819

from = UCA900_WEIGHT_ADDR(dst->weights[page], /*level=*/0, code);

3607

1015819

from_stride = UCA900_DISTANCE_BETWEEN_LEVELS;

3608

1015819

ce_cnt = UCA900_NUM_OF_CE(dst->weights[page], code);

}

}

1034642

for (int weight_ind = 0;

3613

3/4

✓ Branch 0 taken 3741231 times.

✓ Branch 1 taken 1034642 times.

✓ Branch 2 taken 3741231 times.

✗ Branch 3 not taken.

4775873

weight_ind < ce_cnt * MY_UCA_900_CE_SIZE && count < to_length;

3614

weight_ind++) {

3615

3741231

*to = *from;

3616

3741231

to += to_stride;

3617

3741231

from += from_stride;

3618

3741231

count++;

3619

}

3620

1034642

total_ce_cnt += ce_cnt;

}

/*

For shift on primary weight, there might be no enough room in the tables.

3625

For example, Sihala has the rule "&\\u0DA5 < \\u0DA4", which means

3626

that we should move U+0DA4 after U+0DA5 (on the primary level).

3627

However, there is no room after U+0DA5 in DUCET unless we wanted to

3628

conflict with U+0DA6:

3629

3630

0DA4 ; [.28EC.0020.0002] # SINHALA LETTER TAALUJA NAASIKYAYA

3631

0DA5 ; [.28ED.0020.0002] # SINHALA LETTER TAALUJA SANYOOGA NAAKSIKYAYA

3632

0DA6 ; [.28EE.0020.0002] # SINHALA LETTER SANYAKA JAYANNA

3633

3634

Before our implementation of UCA 9.0.0, the shift on primary weight was

3635

done by making it a fake expansion when parsing the rule, where we'd expand

3636

U+0DA4 to U+0DA5 U+MMMM, MMMM being 'last_non_ignorable'. (This happens

3637

in my_coll_parser_scan_reset_sequence()). But from UCA 9.0.0, we also

3638

support accent- and case-sensitive collations, and then, having the extra

3639

weights of 'last_non_ignorable' (which is just a random character) on the

3640

second and third level may cause unexpected results for algorithms that

3641

use the meaning of the tertiary weight to infer case. Thus, we'll abandon

3642

the fake expansion way; instead, instead add an extra CE (after the one

3643

from U+0DA5, the character we are moving after) to represent all the

3644

weights we might want to shift. The actual shifting happens in

3645

apply_shift_900().

3646

3647

For the rule "&\\u0DA5 < \\u0DA4", U+0DA4's weights become

3648

[.28ED.0020.0002][.54A4.0000.0000], where 0x54A4 is the value of

3649

extra_ce_pri_base. We then apply the differences from the rule

3650

(which are never negative) to the last CE, so that it becomes

3651

e.g. [.54A5.0000.0000].

3652

*/

3653

7/8

✓ Branch 0 taken 617612 times.

✓ Branch 1 taken 117155 times.

✓ Branch 2 taken 584834 times.

✓ Branch 3 taken 32778 times.

✓ Branch 4 taken 287768 times.

✓ Branch 5 taken 297066 times.

✓ Branch 6 taken 437701 times.

✗ Branch 7 not taken.

734767

if ((rule->diff[0] || rule->diff[1] || rule->diff[2]) && count < to_length) {

3654

2/2

✓ Branch 0 taken 117155 times.

✓ Branch 1 taken 320546 times.

437701

*to = rule->diff[0] ? dst->extra_ce_pri_base : 0;

3655

437701

to += to_stride;

3656

2/2

✓ Branch 0 taken 46742 times.

✓ Branch 1 taken 390959 times.

437701

*to = rule->diff[1] ? dst->extra_ce_sec_base : 0;

3657

437701

to += to_stride;

3658

2/2

✓ Branch 0 taken 371062 times.

✓ Branch 1 taken 66639 times.

437701

*to = rule->diff[2] ? dst->extra_ce_ter_base : 0;

3659

437701

to += to_stride;

3660

437701

total_ce_cnt++;

3661

437701

count += 3;

3662

}

3663

734767

total_ce_cnt =

3664

734767

std::min(total_ce_cnt, (MY_UCA_MAX_WEIGHT_SIZE - 1) / MY_UCA_900_CE_SIZE);

3665

734767

*to_num_ce = total_ce_cnt;

3666

3667

734767

return total_ce_cnt;

}

/**

Helper function:

Copies UCA weights for a given "uint" string

3673

to the given location.

3674

3675

@param dst destination UCA weight data

3676

@param to destination address

3677

@param to_stride number of bytes between each successive weight in "to"

3678

@param to_length size of destination

3679

@param to_num_ce where to put the number of CEs generated

3680

@param rule The rule that contains the characters whose weight

3681

are to copied

3682

@param base_len The length of base character list

3683

@param uca_ver UCA version

3684

3685

@return number of weights put

3686

*/

3687

3688

1250794

static size_t my_char_weight_put(MY_UCA_INFO *dst, uint16 *to, size_t to_stride,

3689

size_t to_length, uint16 *to_num_ce,

3690

const MY_COLL_RULE *rule, size_t base_len,

3691

enum_uca_ver uca_ver) {

3692

2/2

✓ Branch 0 taken 734767 times.

✓ Branch 1 taken 516027 times.

1250794

if (uca_ver == UCA_V900)

3693

734767

return my_char_weight_put_900(dst, to, to_stride, to_length, to_num_ce,

3694

734767

rule, base_len);

3695

3696

516027

const my_wc_t *base = rule->base;

3697

516027

size_t count = 0;

3698

2/2

✓ Branch 0 taken 525319 times.

✓ Branch 1 taken 516027 times.

1041346

while (base_len != 0) {

3699

525319

const uint16 *from = nullptr;

3700

3701

2/2

✓ Branch 0 taken 10608 times.

✓ Branch 1 taken 524115 times.

534723

for (size_t chlen = base_len; chlen > 1; chlen--) {

3702

2/2

✓ Branch 0 taken 1204 times.

✓ Branch 1 taken 9404 times.

10608

if ((from = my_uca_contraction_weight(dst->contraction_nodes, base,

3703

chlen))) {

3704

1204

base += chlen;

3705

1204

base_len -= chlen;

3706

1204

break;

}

}

2/2

✓ Branch 0 taken 524115 times.

✓ Branch 1 taken 1204 times.

525319

if (!from) {

3711

524115

from = my_char_weight_addr(dst, *base);

3712

524115

base++;

3713

524115

base_len--;

3714

}

3715

3716

4/6

✓ Branch 0 taken 1047646 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 522327 times.

✓ Branch 3 taken 525319 times.

✓ Branch 4 taken 522327 times.

✗ Branch 5 not taken.

1047646

for (; from && *from && count < to_length;) {

3717

522327

*to = *from++;

3718

522327

to += to_stride;

3719

522327

count++;

}

}

516027

*to = 0;

3724

516027

return count;

}

/**

Alloc new page and copy the default UCA weights

3729

@param cs Character set

3730

@param loader Character set loader

3731

@param src Default UCA data to copy from

3732

@param dst UCA data to copy weights to

3733

@param page page number

3734

3735

@retval false on success

3736

@retval true on error

3737

*/

3738

81306

static bool my_uca_copy_page(CHARSET_INFO *cs, MY_CHARSET_LOADER *loader,

3739

const MY_UCA_INFO *src, MY_UCA_INFO *dst,

3740

size_t page) {

3741

81306

const uint dst_size = 256 * dst->lengths[page] * sizeof(uint16);

3742

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 81306 times.

81306

if (!(dst->weights[page] = (uint16 *)(loader->once_alloc)(dst_size)))

3743

✗

return true;

3744

3745

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 81306 times.

81306

assert(src->lengths[page] <= dst->lengths[page]);

3746

81306

memset(dst->weights[page], 0, dst_size);

3747

3/4

✓ Branch 0 taken 81306 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 50238 times.

✓ Branch 3 taken 31068 times.

81306

if (cs->uca && cs->uca->version == UCA_V900) {

3748

50238

const uint src_size = 256 * src->lengths[page] * sizeof(uint16);

3749

50238

memcpy(dst->weights[page], src->weights[page], src_size);

3750

2/2

✓ Branch 0 taken 30734 times.

✓ Branch 1 taken 334 times.

81306

} else if (src->lengths[page] > 0) {

3751

2/2

✓ Branch 0 taken 7867904 times.

✓ Branch 1 taken 30734 times.

7898638

for (uint chc = 0; chc < 256; chc++) {

3752

7867904

memcpy(dst->weights[page] + chc * dst->lengths[page],

3753

7867904

src->weights[page] + chc * src->lengths[page],

3754

7867904

src->lengths[page] * sizeof(uint16));

3755

}

3756

}

3757

81306

return false;

}

/*

This is used to apply the weight shift if there is a [before 1] rule.

3762

If we have a rule "&[before 1] A < B < C", and A's collation element is [P, S,

3763

T], then in my_char_weight_put_900(), we append one extra collation element to

3764

A's CE to be B and C's CE. So B and C's CE becomes [P, S, T][p, 0, 0]. What we

3765

do with this function is to change B's CE to [P - 1, S, T][p + n, 0, 0].

3766

1. The rule "&[before 1] A < B < C" means "B < C < A" on primary level. Since

3767

"B < A", so we give B the first primary weight as (P - 1).

3768

2. p is a weight value which is the maximum regular primary weight in DUCET

3769

plus one (0x54A3 + 1 = 0x54A4). This is to make sure B's primary weight

3770

less than A and greater than any character which sorts before A.

3771

3. n is the number of characters in this rule's character list. For the B in

3772

this rule, n = 1. For the C in this rule, n = 2. This can make sure "B <

3773

C".

3774

3775

It is the same thing that apply_secondary_shift_900() and

3776

apply_tertiary_shift_900() do, but on different weight levels.

3777

*/

3778

38970

static bool apply_primary_shift_900(MY_CHARSET_LOADER *loader,

3779

MY_COLL_RULES *rules, MY_COLL_RULE *r,

3780

uint16 *to, size_t to_stride,

3781

size_t nweights,

3782

uint16 *const last_weight_ptr) {

3783

/*

3784

Find the second-to-last non-ignorable primary weight to apply shift,

3785

because the last one is the extra CE we added in my_char_weight_put_900().

3786

*/

3787

38970

int last_sec_pri = 0;

3788

1/2

✓ Branch 0 taken 38970 times.

✗ Branch 1 not taken.

38970

for (last_sec_pri = nweights - 2; last_sec_pri >= 0; --last_sec_pri) {

3789

1/2

✓ Branch 0 taken 38970 times.

✗ Branch 1 not taken.

38970

if (to[last_sec_pri * to_stride * MY_UCA_900_CE_SIZE]) break;

3790

}

3791

1/2

✓ Branch 0 taken 38970 times.

✗ Branch 1 not taken.

38970

if (last_sec_pri >= 0) {

3792

38970

to[last_sec_pri * to_stride * MY_UCA_900_CE_SIZE]--; /* Reset before */

3793

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 38970 times.

38970

if (rules->shift_after_method == my_shift_method_expand) {

3794

/*

3795

Special case. Don't let characters shifted after X

3796

and before next(X) intermix to each other.

3797

3798

For example:

3799

"[shift-after-method expand] &0 < a &[before primary]1 < A".

3800

I.e. we reorder 'a' after '0', and then 'A' before '1'.

3801

'a' must be sorted before 'A'.

3802

3803

Note, there are no real collations in CLDR which shift

3804

after and before two neighbouring characters. We need this

3805

just in case. Reserving 4096 (0x1000) weights for such

3806

cases is perfectly enough.

3807

*/

3808

/* W3-TODO: const may vary on levels 2,3*/

3809

✗

last_weight_ptr[0] += 0x1000;

3810

}

3811

} else {

3812

✗

loader->errcode = EE_FAILED_TO_RESET_BEFORE_PRIMARY_IGNORABLE_CHAR;

3813

✗

snprintf(loader->errarg, sizeof(loader->errarg), "U+%04lX", r->base[0]);

3814

✗

return true;

3815

}

3816

38970

return false;

}

/*

This is used to apply the weight shift if there is a [before 2] rule. Please

3821

see the comment on apply_primary_shift_900().

3822

*/

3823

8960

static bool apply_secondary_shift_900(MY_CHARSET_LOADER *loader,

3824

MY_COLL_RULES *rules, MY_COLL_RULE *r,

3825

uint16 *to, size_t to_stride,

3826

size_t nweights,

3827

uint16 *const last_weight_ptr) {

3828

/*

3829

Find the second-to-last non-ignorable secondary weight to apply shift,

3830

because the last one is the extra CE we added in my_char_weight_put_900().

3831

*/

3832

int last_sec_sec;

3833

1/2

✓ Branch 0 taken 8960 times.

✗ Branch 1 not taken.

8960

for (last_sec_sec = nweights - 2; last_sec_sec >= 0; --last_sec_sec) {

3834

1/2

✓ Branch 0 taken 8960 times.

✗ Branch 1 not taken.

8960

if (to[last_sec_sec * MY_UCA_900_CE_SIZE * to_stride + to_stride]) break;

3835

}

3836

1/2

✓ Branch 0 taken 8960 times.

✗ Branch 1 not taken.

8960

if (last_sec_sec >= 0) {

3837

// Reset before.

3838

8960

to[last_sec_sec * MY_UCA_900_CE_SIZE * to_stride + to_stride]--;

3839

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 8960 times.

8960

if (rules->shift_after_method == my_shift_method_expand) {

3840

/*

3841

Same reason as in apply_primary_shift_900(), reserve 256 (0x100)

3842

weights for secondary level.

3843

*/

3844

✗

last_weight_ptr[to_stride] += 0x100;

3845

}

3846

} else {

3847

✗

loader->errcode = EE_FAILED_TO_RESET_BEFORE_SECONDARY_IGNORABLE_CHAR;

3848

✗

snprintf(loader->errarg, sizeof(loader->errarg), "U+%04lX", r->base[0]);

3849

✗

return true;

3850

}

3851

8960

return false;

}

/*

This is used to apply the weight shift if there is a [before 3] rule. Please

3856

see the comment on apply_primary_shift_900().

3857

*/

3858

204459

static bool apply_tertiary_shift_900(MY_CHARSET_LOADER *loader,

3859

MY_COLL_RULES *rules, MY_COLL_RULE *r,

3860

uint16 *to, size_t to_stride,

3861

size_t nweights,

3862

uint16 *const last_weight_ptr) {

3863

/*

3864

Find the second-to-last non-ignorable tertiary weight to apply shift,

3865

because the last one is the extra CE we added in my_char_weight_put_900().

3866

*/

3867

int last_sec_ter;

3868

1/2

✓ Branch 0 taken 204459 times.

✗ Branch 1 not taken.

204459

for (last_sec_ter = nweights - 2; last_sec_ter >= 0; --last_sec_ter) {

3869

1/2

✓ Branch 0 taken 204459 times.

✗ Branch 1 not taken.

204459

if (to[last_sec_ter * MY_UCA_900_CE_SIZE * to_stride + 2 * to_stride])

3870

204459

break;

3871

}

3872

1/2

✓ Branch 0 taken 204459 times.

✗ Branch 1 not taken.

204459

if (last_sec_ter >= 0) {

3873

// Reset before.

3874

204459

to[last_sec_ter * MY_UCA_900_CE_SIZE * to_stride + 2 * to_stride]--;

3875

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 204459 times.

204459

if (rules->shift_after_method == my_shift_method_expand) {

3876

/*

3877

Same reason as in apply_primary_shift_900(), reserve 16 (0x10)

3878

weights for tertiary level.

3879

*/

3880

✗

last_weight_ptr[to_stride * 2] += 0x10;

3881

}

3882

} else {

3883

✗

loader->errcode = EE_FAILED_TO_RESET_BEFORE_TERTIARY_IGNORABLE_CHAR;

3884

✗

snprintf(loader->errarg, sizeof(loader->errarg), "U+%04lX", r->base[0]);

3885

✗

return true;

3886

}

3887

204459

return false;

3888

}

3889

3890

734767

static bool apply_shift_900(MY_CHARSET_LOADER *loader, MY_COLL_RULES *rules,

3891

MY_COLL_RULE *r, uint16 *to, size_t to_stride,

3892

size_t nweights) {

3893

// nweights should not less than 1 because of the extra CE.

3894

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 734767 times.

734767

assert(nweights);

3895

// Apply level difference.

3896

734767

uint16 *const last_weight_ptr =

3897

734767

to + (nweights - 1) * to_stride * MY_UCA_900_CE_SIZE;

3898

734767

last_weight_ptr[0] += r->diff[0];

3899

734767

last_weight_ptr[to_stride] += r->diff[1];

3900

734767

last_weight_ptr[to_stride * 2] += r->diff[2];

3901

2/2

✓ Branch 0 taken 38970 times.

✓ Branch 1 taken 695797 times.

734767

if (r->before_level == 1) // Apply "&[before primary]".

3902

38970

return apply_primary_shift_900(loader, rules, r, to, to_stride, nweights,

3903

38970

last_weight_ptr);

3904

2/2

✓ Branch 0 taken 8960 times.

✓ Branch 1 taken 686837 times.

695797

else if (r->before_level == 2) // Apply "[before 2]".

3905

8960

return apply_secondary_shift_900(loader, rules, r, to, to_stride, nweights,

3906

8960

last_weight_ptr);

3907

2/2

✓ Branch 0 taken 204459 times.

✓ Branch 1 taken 482378 times.

686837

else if (r->before_level == 3) // Apply "[before 3]".

3908

204459

return apply_tertiary_shift_900(loader, rules, r, to, to_stride, nweights,

3909

204459

last_weight_ptr);

3910

482378

return false;

3911

}

3912

3913

1250794

static bool apply_shift(MY_CHARSET_LOADER *loader, MY_COLL_RULES *rules,

3914

MY_COLL_RULE *r, int level, uint16 *to,

3915

size_t to_stride, size_t nweights) {

3916

2/2

✓ Branch 0 taken 734767 times.

✓ Branch 1 taken 516027 times.

1250794

if (rules->uca->version == UCA_V900)

3917

734767

return apply_shift_900(loader, rules, r, to, to_stride, nweights);

3918

3919

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 516027 times.

516027

assert(to_stride == 1);

3920

3921

/* Apply level difference. */

3922

2/2

✓ Branch 0 taken 512747 times.

✓ Branch 1 taken 3280 times.

516027

if (nweights) {

3923

512747

to[nweights - 1] += r->diff[0];

3924

2/2

✓ Branch 0 taken 186 times.

✓ Branch 1 taken 512561 times.

512747

if (r->before_level == 1) /* Apply "&[before primary]" */

3925

{

3926

2/2

✓ Branch 0 taken 184 times.

✓ Branch 1 taken 2 times.

186

if (nweights >= 2) {

3927

184

to[nweights - 2]--; /* Reset before */

3928

2/2

✓ Branch 0 taken 56 times.

✓ Branch 1 taken 128 times.

184

if (rules->shift_after_method == my_shift_method_expand) {

3929

/*

3930

Special case. Don't let characters shifted after X

3931

and before next(X) intermix to each other.

3932

3933

For example:

3934

"[shift-after-method expand] &0 < a &[before primary]1 < A".

3935

I.e. we reorder 'a' after '0', and then 'A' before '1'.

3936

'a' must be sorted before 'A'.

3937

3938

Note, there are no real collations in CLDR which shift

3939

after and before two neighbour characters. We need this

3940

just in case. Reserving 4096 (0x1000) weights for such

3941

cases is perfectly enough.

3942

*/

3943

/* W3-TODO: const may vary on levels 2,3*/

3944

56

to[nweights - 1] += 0x1000;

3945

}

3946

} else {

3947

2

loader->errcode = EE_FAILED_TO_RESET_BEFORE_PRIMARY_IGNORABLE_CHAR;

3948

2

snprintf(loader->errarg, sizeof(loader->errarg), "U+%04lX", r->base[0]);

3949

2

return true;

}

}

} else {

/* Shift to an ignorable character, e.g.: & \u0000 < \u0001 */

3954

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 3280 times.

3280

assert(to[0] == 0);

3955

3280

to[0] = r->diff[level];

3956

}

3957

516025

return false;

3958

}

3959

3960

267155

static MY_CONTRACTION *add_contraction_to_trie(

3961

std::vector<MY_CONTRACTION> *cont_nodes, MY_COLL_RULE *r) {

3962

267155

MY_CONTRACTION new_node{0, {}, {}, {}, false, 0};

3963

2/2

✓ Branch 0 taken 204479 times.

✓ Branch 1 taken 62676 times.

267155

if (r->with_context) // previous-context contraction

3964

{

3965

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 204479 times.

204479

assert(my_wstrnlen(r->curr, MY_UCA_MAX_CONTRACTION) == 2);

3966

std::vector<MY_CONTRACTION>::iterator node_it =

3967

1/2

✓ Branch 0 taken 204479 times.

✗ Branch 1 not taken.

204479

find_contraction_part_in_trie(*cont_nodes, r->curr[1]);

3968

6/6

✓ Branch 0 taken 203510 times.

✓ Branch 1 taken 969 times.

✓ Branch 2 taken 650 times.

✓ Branch 3 taken 202860 times.

✓ Branch 4 taken 1619 times.

✓ Branch 5 taken 202860 times.

204479

if (node_it == cont_nodes->end() || node_it->ch != r->curr[1]) {

3969

1619

new_node.ch = r->curr[1];

3970

1/2

✓ Branch 0 taken 1619 times.

✗ Branch 1 not taken.

1619

node_it = cont_nodes->insert(node_it, new_node);

3971

}

3972

204479

cont_nodes = &node_it->child_nodes_context;

3973

3974

1/2

✓ Branch 0 taken 204479 times.

✗ Branch 1 not taken.

204479

node_it = find_contraction_part_in_trie(*cont_nodes, r->curr[0]);

3975

6/6

✓ Branch 0 taken 175066 times.

✓ Branch 1 taken 29413 times.

✓ Branch 2 taken 173128 times.

✓ Branch 3 taken 1938 times.

✓ Branch 4 taken 202541 times.

✓ Branch 5 taken 1938 times.

204479

if (node_it == cont_nodes->end() || node_it->ch != r->curr[0]) {

3976

202541

new_node.ch = r->curr[0];

3977

1/2

✓ Branch 0 taken 202541 times.

✗ Branch 1 not taken.

202541

node_it = cont_nodes->insert(node_it, new_node);

3978

}

3979

204479

node_it->is_contraction_tail = true;

3980

204479

node_it->contraction_len = 2;

3981

204479

return &(*node_it);

3982

} else // normal contraction

3983

{

3984

62676

size_t contraction_len = my_wstrnlen(r->curr, MY_UCA_MAX_CONTRACTION);

3985

62676

std::vector<MY_CONTRACTION>::iterator node_it;

3986

2/2

✓ Branch 0 taken 137833 times.

✓ Branch 1 taken 62676 times.

200509

for (size_t ch_ind = 0; ch_ind < contraction_len; ++ch_ind) {

3987

1/2

✓ Branch 0 taken 137833 times.

✗ Branch 1 not taken.

137833

node_it = find_contraction_part_in_trie(*cont_nodes, r->curr[ch_ind]);

3988

6/6

✓ Branch 0 taken 77932 times.

✓ Branch 1 taken 59901 times.

✓ Branch 2 taken 42511 times.

✓ Branch 3 taken 35421 times.

✓ Branch 4 taken 102412 times.

✓ Branch 5 taken 35421 times.

137833

if (node_it == cont_nodes->end() || node_it->ch != r->curr[ch_ind]) {

3989

102412

new_node.ch = r->curr[ch_ind];

3990

1/2

✓ Branch 0 taken 102412 times.

✗ Branch 1 not taken.

102412

node_it = cont_nodes->insert(node_it, new_node);

3991

}

3992

137833

cont_nodes = &node_it->child_nodes;

3993

}

3994

62676

node_it->is_contraction_tail = true;

3995

62676

node_it->contraction_len = contraction_len;

3996

62676

return &(*node_it);

3997

}

3998

267155

}

3999

4000

1250794

static bool apply_one_rule(CHARSET_INFO *cs, MY_CHARSET_LOADER *loader,

4001

MY_COLL_RULES *rules, MY_COLL_RULE *r, int level,

4002

MY_UCA_INFO *dst) {

4003

size_t nweights;

4004

1250794

size_t nreset = my_coll_rule_reset_length(r); /* Length of reset sequence */

4005

1250794

size_t nshift = my_coll_rule_shift_length(r); /* Length of shift sequence */

4006

uint16 *to, *to_num_ce;

4007

size_t to_stride;

4008

4009

2/2

✓ Branch 0 taken 267155 times.

✓ Branch 1 taken 983639 times.

1250794

if (nshift >= 2) /* Contraction */

{

size_t i;

int flag;

/* Add HEAD, MID and TAIL flags for the contraction parts */

4014

267155

my_uca_add_contraction_flag(

4015

dst->contraction_flags, r->curr[0],

4016

2/2

✓ Branch 0 taken 204479 times.

✓ Branch 1 taken 62676 times.

267155

r->with_context ? MY_UCA_PREVIOUS_CONTEXT_HEAD : MY_UCA_CNT_HEAD);

4017

2/2

✓ Branch 0 taken 12481 times.

✓ Branch 1 taken 267155 times.

279636

for (i = 1, flag = MY_UCA_CNT_MID1; i < nshift - 1; i++, flag <<= 1)

4018

12481

my_uca_add_contraction_flag(dst->contraction_flags, r->curr[i], flag);

4019

267155

my_uca_add_contraction_flag(

4020

dst->contraction_flags, r->curr[i],

4021

2/2

✓ Branch 0 taken 204479 times.

✓ Branch 1 taken 62676 times.

267155

r->with_context ? MY_UCA_PREVIOUS_CONTEXT_TAIL : MY_UCA_CNT_TAIL);

4022

/* Add new contraction to the contraction list */

4023

MY_CONTRACTION *trie_node =

4024

267155

add_contraction_to_trie(dst->contraction_nodes, r);

4025

267155

to = trie_node->weight;

4026

267155

to_stride = 1;

4027

267155

to_num_ce = &to[MY_UCA_MAX_WEIGHT_SIZE - 1];

4028

/* Store weights of the "reset to" character */

4029

nweights =

4030

267155

my_char_weight_put(dst, to, to_stride, MY_UCA_MAX_WEIGHT_SIZE - 1,

4031

267155

to_num_ce, r, nreset, rules->uca->version);

4032

} else {

4033

983639

my_wc_t pagec = (r->curr[0] >> 8);

4034

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 983639 times.

983639

assert(dst->weights[pagec]);

4035

3/4

✓ Branch 0 taken 983639 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 493240 times.

✓ Branch 3 taken 490399 times.

983639

if (cs->uca && cs->uca->version == UCA_V900) {

4036

493240

to = my_char_weight_addr_900(dst, r->curr[0]);

4037

493240

to_stride = UCA900_DISTANCE_BETWEEN_LEVELS;

4038

493240

to_num_ce = to - UCA900_DISTANCE_BETWEEN_LEVELS;

4039

} else {

4040

490399

to = my_char_weight_addr(dst, r->curr[0]);

4041

490399

to_stride = 1;

4042

490399

to_num_ce = to + (dst->lengths[pagec] - 1);

4043

}

4044

/* Store weights of the "reset to" character */

4045

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 983639 times.

983639

if (dst->lengths[pagec] == 0)

4046

✗

nweights = 0;

4047

else

4048

983639

nweights = my_char_weight_put(dst, to, to_stride, dst->lengths[pagec] - 1,

4049

983639

to_num_ce, r, nreset, rules->uca->version);

4050

}

4051

4052

1250794

change_weight_if_case_first(cs, dst, r, to, to_stride, nshift, nweights);

4053

/* Apply level difference. */

4054

1250794

return apply_shift(loader, rules, r, level, to, to_stride, nweights);

}

/**

Check if collation rules are valid,

4059

i.e. characters are not outside of the collation supported range.

4060

*/

4061

29006

static int check_rules(MY_CHARSET_LOADER *loader, const MY_COLL_RULES *rules,

4062

const MY_UCA_INFO *dst, const MY_UCA_INFO *src) {

4063

const MY_COLL_RULE *r, *rlast;

4064

2/2

✓ Branch 0 taken 1250794 times.

✓ Branch 1 taken 29006 times.

1279800

for (r = rules->rule, rlast = rules->rule + rules->nrules; r < rlast; r++) {

4065

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1250794 times.

1250794

if (r->curr[0] > dst->maxchar) {

4066

✗

loader->errcode = EE_SHIFT_CHAR_OUT_OF_RANGE;

4067

✗

snprintf(loader->errarg, sizeof(loader->errarg), "u%04X",

4068

✗

(uint)r->curr[0]);

4069

✗

return true;

4070

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 1250794 times.

1250794

} else if (r->base[0] > src->maxchar) {

4071

✗

loader->errcode = EE_RESET_CHAR_OUT_OF_RANGE;

4072

✗

snprintf(loader->errarg, sizeof(loader->errarg), "u%04X",

4073

✗

(uint)r->base[0]);

4074

✗

return true;

4075

}

4076

}

4077

29006

return false;

4078

}

4079

4080

9359

static void synthesize_lengths_900(uchar *lengths, const uint16 *const *weights,

4081

uint npages) {

4082

2/2

✓ Branch 0 taken 40730368 times.

✓ Branch 1 taken 9359 times.

40739727

for (uint page = 0; page < npages; ++page) {

4083

40730368

int max_len = 0;

4084

2/2

✓ Branch 0 taken 1396105 times.

✓ Branch 1 taken 39334263 times.

40730368

if (weights[page]) {

4085

2/2

✓ Branch 0 taken 357402880 times.

✓ Branch 1 taken 1396105 times.

358798985

for (uint code = 0; code < 256; ++code) {

4086

357402880

max_len = std::max<int>(max_len, weights[page][code]);

4087

}

4088

}

4089

2/2

✓ Branch 0 taken 39334263 times.

✓ Branch 1 taken 1396105 times.

40730368

if (max_len == 0)

4090

39334263

lengths[page] = 0;

4091

else

4092

1396105

lengths[page] = max_len * MY_UCA_900_CE_SIZE + 1;

4093

}

4094

9359

}

4095

4096

28846

static void copy_ja_han_pages(const CHARSET_INFO *cs, MY_UCA_INFO *dst) {

4097

3/4

✓ Branch 0 taken 28846 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 9199 times.

✓ Branch 3 taken 19647 times.

28846

if (!cs->uca || cs->uca->version != UCA_V900 ||

4098

2/2

✓ Branch 0 taken 8876 times.

✓ Branch 1 taken 323 times.

9199

cs->coll_param != &ja_coll_param)

4099

28523

return;

4100

2/2

✓ Branch 0 taken 26486 times.

✓ Branch 1 taken 323 times.

26809

for (int page = MIN_JA_HAN_PAGE; page <= MAX_JA_HAN_PAGE; page++) {

4101

// In DUCET, weight is not assigned to code points in [U+4E00, U+9FFF].

4102

// When re-initializing (after my_coll_uninit_uca), the weights

4103

// may already be set.

4104

3/4

✓ Branch 0 taken 738 times.

✓ Branch 1 taken 25748 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 738 times.

26486

assert(dst->weights[page] == nullptr ||

4105

dst->weights[page] == ja_han_pages[page - MIN_JA_HAN_PAGE]);

4106

26486

dst->weights[page] = ja_han_pages[page - MIN_JA_HAN_PAGE];

}

}

/*

We have reordered all the characters in the pages which contains Chinese Han

4112

characters with uca9dump (see dump_zh_pages() in uca9-dump.cc). Replace the

4113

DUCET pages with these pages.

4114

*/

4115

160

static void copy_zh_han_pages(MY_UCA_INFO *dst) {

4116

2/2

✓ Branch 0 taken 107680 times.

✓ Branch 1 taken 160 times.

107840

for (int page = MIN_ZH_HAN_PAGE; page <= MAX_ZH_HAN_PAGE; page++) {

4117

2/2

✓ Branch 0 taken 47200 times.

✓ Branch 1 taken 60480 times.

107680

if (zh_han_pages[page - MIN_ZH_HAN_PAGE]) {

4118

47200

dst->weights[page] = zh_han_pages[page - MIN_ZH_HAN_PAGE];

4119

}

4120

}

4121

160

}

4122

4123

/*

4124

UCA defines an algorithm to calculate character's implicit weight if this

4125

character's weight is not defined in the DUCET. This function is to help

4126

convert Chinese character's implicit weight calculated by UCA back to its code

4127

points.

4128

The implicit weight and the code point is not 1 : 1 map because DUCET lets

4129

some characters share implicit primary weight. For example, the DUCET defines

4130

"2F00 ; [.FB40.0020.0004][.CE00.0000.0000] # KANGXI RADICAL ONE", and 4E00's

4131

implicit weight is [.FB40.0020.0002][.CE00.0000.0000]. We can see the primary

4132

weights of U+2F00 and U+4E00 are same (FB40 CE00).

4133

4134

But for the Han characters in zh.xml file, each one has unique implicit

4135

weight.

4136

*/

4137

498638

static inline my_wc_t convert_implicit_to_ch(uint16 first, uint16 second) {

4138

/*

4139

For reference, here is how UCA calculates one character's implicit weight.

4140

AAAA = 0xFB40 + (CP >> 15) # The 0xFB40 changes for different character

4141

# groups

4142

BBBB = (CP & 0x7FFF) | 0x8000

4143

*/

4144

2/2

✓ Branch 0 taken 201280 times.

✓ Branch 1 taken 297358 times.

498638

if (first < 0xFB80)

4145

201280

return (((first - 0xFB40) << 15) | (second & 0x7FFF));

4146

2/2

✓ Branch 0 taken 28574 times.

✓ Branch 1 taken 268784 times.

297358

else if (first < 0xFBC0)

4147

28574

return (((first - 0xFB80) << 15) | (second & 0x7FFF));

4148

else

4149

268784

return (((first - 0xFBC0) << 15) | (second & 0x7FFF));

}

/*

Usually we do reordering in apply_reorder_param(). But for the Chinese

4154

collation, since we want to remove the weight gap between the character groups

4155

(see the comment on change_zh_implicit()), and we have done the reordering for

4156

some characters in the pages which contains Chinese Han characters, if we

4157

still use apply_reorder_param() to do the reordering for other characters, we

4158

might meet weight conflict. For example, in the DUCET page, 'A' has primary

4159

weight 0x1C47, but this value has been assigned to the first Chinese Han

4160

character in CLDR zh.xml file.

4161

So we do the reordering for all the DUCET pages when initializing the

4162

collation.

4163

*/

4164

160

static void modify_all_zh_pages(Reorder_param *reorder_param, MY_UCA_INFO *dst,

4165

int npages) {

4166

160

std::map<int, int> zh_han_to_single_weight_map;

4167

2/2

✓ Branch 0 taken 6613760 times.

✓ Branch 1 taken 160 times.

6613920

for (int i = 0; i < ZH_HAN_WEIGHT_PAIRS; i++) {

4168

6613760

zh_han_to_single_weight_map[zh_han_to_single_weight[i * 2]] =

4169

1/2

✓ Branch 0 taken 6613760 times.

✗ Branch 1 not taken.

6613760

zh_han_to_single_weight[i * 2 + 1];

4170

}

4171

4172

2/2

✓ Branch 0 taken 696320 times.

✓ Branch 1 taken 160 times.

696480

for (int page = 0; page < npages; page++) {

4173

/*

4174

If there is no page in the DUCET, then all the characters in this page

4175

must have implicit weight. The reordering for it will be done by

4176

change_zh_implicit(). Do not need to change here.

4177

If there is page in zh_han_pages[], then all the characters in this page

4178

have been reordered by uca9dump. Do not need to change here.

4179

*/

4180

2/2

✓ Branch 0 taken 24716 times.

✓ Branch 1 taken 671604 times.

696320

if (!dst->weights[page] ||

4181

4/4

✓ Branch 0 taken 17356 times.

✓ Branch 1 taken 7360 times.

✓ Branch 2 taken 16556 times.

✓ Branch 3 taken 800 times.

24716

(page >= MIN_ZH_HAN_PAGE && page <= MAX_ZH_HAN_PAGE &&

4182

2/2

✓ Branch 0 taken 1356 times.

✓ Branch 1 taken 15200 times.

16556

zh_han_pages[page - MIN_ZH_HAN_PAGE]))

4183

672960

continue;

4184

2/2

✓ Branch 0 taken 5980160 times.

✓ Branch 1 taken 23360 times.

6003520

for (int off = 0; off < 256; off++) {

4185

5980160

uint16 *wbeg = UCA900_WEIGHT_ADDR(dst->weights[page], 0, off);

4186

5980160

int num_of_ce = UCA900_NUM_OF_CE(dst->weights[page], off);

4187

2/2

✓ Branch 0 taken 6612156 times.

✓ Branch 1 taken 5980160 times.

12592316

for (int ce = 0; ce < num_of_ce; ce++) {

4188

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 6612156 times.

6612156

assert(reorder_param->wt_rec_num == 1);

4189

2/2

✓ Branch 0 taken 4781389 times.

✓ Branch 1 taken 1830767 times.

6612156

if (*wbeg >= reorder_param->wt_rec[0].old_wt_bdy.begin &&

4190

2/2

✓ Branch 0 taken 3200440 times.

✓ Branch 1 taken 1580949 times.

4781389

*wbeg <= reorder_param->wt_rec[0].old_wt_bdy.end) {

4191

3200440

*wbeg = *wbeg + reorder_param->wt_rec[0].new_wt_bdy.begin -

4192

3200440

reorder_param->wt_rec[0].old_wt_bdy.begin;

4193

2/2

✓ Branch 0 taken 1469469 times.

✓ Branch 1 taken 1942247 times.

3411716

} else if (*wbeg >= 0xFB00) {

4194

1469469

uint16 next_wt = *(wbeg + UCA900_DISTANCE_BETWEEN_WEIGHTS);

4195

4/4

✓ Branch 0 taken 1469409 times.

✓ Branch 1 taken 60 times.

✓ Branch 2 taken 498638 times.

✓ Branch 3 taken 970771 times.

1469469

if (*wbeg >= 0xFB40 && *wbeg <= 0xFBC1) { // Han's implicit weight

4196

/*

4197

If some characters in DUCET share the same implicit weight, their

4198

reordered weight should be same too.

4199

*/

4200

498638

my_wc_t ch = convert_implicit_to_ch(*wbeg, next_wt);

4201

1/2

✓ Branch 0 taken 498638 times.

✗ Branch 1 not taken.

498638

if (zh_han_to_single_weight_map.find(ch) !=

4202

2/2

✓ Branch 0 taken 215721 times.

✓ Branch 1 taken 282917 times.

997276

zh_han_to_single_weight_map.end()) {

4203

1/2

✓ Branch 0 taken 215721 times.

✗ Branch 1 not taken.

215721

*wbeg = zh_han_to_single_weight_map[ch];

4204

215721

*(wbeg + UCA900_DISTANCE_BETWEEN_WEIGHTS) = 0;

4205

215721

wbeg += UCA900_DISTANCE_BETWEEN_WEIGHTS;

4206

215721

ce++;

4207

215721

continue;

4208

}

4209

}

4210

1253748

*wbeg = change_zh_implicit(*wbeg);

4211

1253748

wbeg += UCA900_DISTANCE_BETWEEN_WEIGHTS;

4212

1253748

ce++;

4213

}

4214

6396435

wbeg += UCA900_DISTANCE_BETWEEN_WEIGHTS;

}

}

}

160

}

4219

4220

29006

static bool init_weight_level(CHARSET_INFO *cs, MY_CHARSET_LOADER *loader,

4221

MY_COLL_RULES *rules, int level, MY_UCA_INFO *dst,

4222

const MY_UCA_INFO *src,

4223

bool lengths_are_temporary) {

4224

MY_COLL_RULE *r, *rlast;

4225

29006

size_t i, npages = (src->maxchar + 1) / 256;

4226

29006

bool has_contractions = false;

4227

4228

29006

dst->maxchar = src->maxchar;

4229

4230

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 29006 times.

29006

if (check_rules(loader, rules, dst, src)) return true;

4231

4232

/* Allocate memory for pages and their lengths */

4233

2/2

✓ Branch 0 taken 9359 times.

✓ Branch 1 taken 19647 times.

29006

if (lengths_are_temporary) {

4234

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 9359 times.

9359

if (!(dst->lengths = (uchar *)(loader->mem_malloc)(npages))) return true;

4235

9359

if (!(dst->weights =

4236

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 9359 times.

9359

(uint16 **)(loader->once_alloc)(npages * sizeof(uint16 *)))) {

4237

✗

(loader->mem_free)(dst->lengths);

4238

✗

return true;

4239

}

4240

} else {

4241

2/4

✓ Branch 0 taken 19647 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 19647 times.

39294

if (!(dst->lengths = (uchar *)(loader->once_alloc)(npages)) ||

4242

19647

!(dst->weights =

4243

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 19647 times.

19647

(uint16 **)(loader->once_alloc)(npages * sizeof(uint16 *))))

4244

✗

return true;

}

/*

Copy pages lengths and page pointers from the default UCA weights.

4249

*/

4250

29006

memcpy(dst->lengths, src->lengths, npages);

4251

29006

memcpy(dst->weights, src->weights, npages * sizeof(uint16 *));

4252

4253

/*

4254

Calculate maximum lengths for the pages which will be overwritten.

4255

Mark pages that will be overwritten as NULL.

4256

We'll allocate their own memory.

4257

*/

4258

2/2

✓ Branch 0 taken 1250794 times.

✓ Branch 1 taken 29006 times.

1279800

for (r = rules->rule, rlast = rules->rule + rules->nrules; r < rlast; r++) {

4259

2/2

✓ Branch 0 taken 983639 times.

✓ Branch 1 taken 267155 times.

1250794

if (!r->curr[1]) /* If not a contraction */

4260

{

4261

983639

uint pagec = (r->curr[0] >> 8);

4262

2/2

✓ Branch 0 taken 218773 times.

✓ Branch 1 taken 764866 times.

983639

if (r->base[1]) /* Expansion */

4263

{

4264

/* Reserve space for maximum possible length */

4265

218773

dst->lengths[pagec] = MY_UCA_MAX_WEIGHT_SIZE;

4266

} else {

4267

764866

uint pageb = (r->base[0] >> 8);

4268

6/6

✓ Branch 0 taken 251626 times.

✓ Branch 1 taken 513240 times.

✓ Branch 2 taken 162963 times.

✓ Branch 3 taken 88663 times.

✓ Branch 4 taken 53292 times.

✓ Branch 5 taken 109671 times.

764866

if ((r->diff[0] || r->diff[1] || r->diff[2]) &&

4269

2/2

✓ Branch 0 taken 37200 times.

✓ Branch 1 taken 617995 times.

655195

dst->lengths[pagec] < (src->lengths[pageb] + 3)) {

4270

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 37200 times.

37200

if ((src->lengths[pageb] + 3) > MY_UCA_MAX_WEIGHT_SIZE)

4271

✗

dst->lengths[pagec] = MY_UCA_MAX_WEIGHT_SIZE;

4272

else

4273

37200

dst->lengths[pagec] = src->lengths[pageb] + 3;

4274

2/2

✓ Branch 0 taken 648 times.

✓ Branch 1 taken 727018 times.

727666

} else if (dst->lengths[pagec] < src->lengths[pageb])

4275

648

dst->lengths[pagec] = src->lengths[pageb];

4276

}

4277

983639

dst->weights[pagec] = nullptr; /* Mark that we'll overwrite this page */

4278

} else

4279

267155

has_contractions = true;

4280

}

4281

4282

2/2

✓ Branch 0 taken 8844 times.

✓ Branch 1 taken 20162 times.

29006

if (has_contractions) {

4283

8844

dst->have_contractions = true;

4284

2/4

✓ Branch 0 taken 8844 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 8844 times.

✗ Branch 3 not taken.

8844

dst->contraction_nodes = new std::vector<MY_CONTRACTION>(0);

4285

8844

if (!(dst->contraction_flags =

4286

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 8844 times.

8844

(char *)(loader->once_alloc)(MY_UCA_CNT_FLAG_SIZE)))

4287

✗

return true;

4288

8844

memset(dst->contraction_flags, 0, MY_UCA_CNT_FLAG_SIZE);

4289

}

4290

2/2

✓ Branch 0 taken 160 times.

✓ Branch 1 taken 28846 times.

29006

if (cs->coll_param == &zh_coll_param) {

4291

/*

4292

We are going to reorder the weight of characters in uca pages when

4293

initializing this collation. And because of the reorder rule [reorder

4294

Hani], we need to change almost every character's weight. So copy all

4295

the pages.

4296

Please also see the comment on modify_all_zh_pages().

4297

*/

4298

bool rc;

4299

2/2

✓ Branch 0 taken 696320 times.

✓ Branch 1 taken 160 times.

696480

for (i = 0; i < npages; i++) {

4300

4/6

✓ Branch 0 taken 24716 times.

✓ Branch 1 taken 671604 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 24716 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 696320 times.

696320

if (dst->lengths[i] && (rc = my_uca_copy_page(cs, loader, src, dst, i)))

4301

✗

return rc;

4302

}

4303

160

modify_all_zh_pages(cs->coll_param->reorder_param, dst, npages);

4304

160

copy_zh_han_pages(dst);

4305

} else {

4306

/* Allocate pages that we'll overwrite and copy default weights */

4307

2/2

✓ Branch 0 taken 49044992 times.

✓ Branch 1 taken 28846 times.

49073838

for (i = 0; i < npages; i++) {

4308

bool rc;

4309

/*

4310

Don't touch pages with lengths[i]==0, they have implicit weights

4311

calculated algorithmically.

4312

*/

4313

5/6

✓ Branch 0 taken 46520142 times.

✓ Branch 1 taken 2524850 times.

✓ Branch 2 taken 56590 times.

✓ Branch 3 taken 46463552 times.

✗ Branch 4 not taken.

✓ Branch 5 taken 49044992 times.

49101582

if (!dst->weights[i] && dst->lengths[i] &&

4314

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 56590 times.

56590

(rc = my_uca_copy_page(cs, loader, src, dst, i)))

4315

✗

return rc;

4316

}

4317

4318

28846

copy_ja_han_pages(cs, dst);

}

/*

Preparatory step is done at this point.

4323

Now we have memory allocated for the pages that we'll overwrite,

4324

and for contractions, including previous context contractions.

4325

Also, for the pages that we'll overwrite, we have copied default weights.

4326

Now iterate through the rules, overwrite weights for the characters

4327

that appear in the rules, and put all contractions into contraction list.

4328

*/

4329

2/2

✓ Branch 0 taken 1250794 times.

✓ Branch 1 taken 29004 times.

1279798

for (r = rules->rule; r < rlast; r++) {

4330

2/2

✓ Branch 0 taken 2 times.

✓ Branch 1 taken 1250792 times.

1250794

if (apply_one_rule(cs, loader, rules, r, level, dst)) return true;

4331

}

4332

29004

return false;

}

/**

Check whether the composition character is already in rule list

4337

@param rules The rule list

4338

@param wc The composition character

4339

@return true The composition character is already in list

4340

false The composition character is not in list

4341

*/

4342

578607

static bool my_comp_in_rulelist(const MY_COLL_RULES *rules, my_wc_t wc) {

4343

MY_COLL_RULE *r, *rlast;

4344

2/2

✓ Branch 0 taken 100543160 times.

✓ Branch 1 taken 468377 times.

101011537

for (r = rules->rule, rlast = rules->rule + rules->nrules; r < rlast; r++) {

4345

4/4

✓ Branch 0 taken 138654 times.

✓ Branch 1 taken 100404506 times.

✓ Branch 2 taken 110230 times.

✓ Branch 3 taken 28424 times.

100543160

if (r->curr[0] == wc && r->curr[1] == 0) return true;

4346

}

4347

468377

return false;

}

/**

Check whether a composition character in the decomposition list is a

4352

normal character.

4353

@param dec_ind The index of composition character in list

4354

@return Whether it is a normal character

4355

*/

4356

1761683638

static inline bool my_compchar_is_normal_char(uint dec_ind) {

4357

1761683638

return uni_dec[dec_ind].decomp_tag == DECOMP_TAG_NONE;

4358

}

4359

4360

904076

static inline bool my_compchar_is_normal_char(const Unidata_decomp *decomp) {

4361

904076

return my_compchar_is_normal_char(decomp - std::begin(uni_dec));

4362

}

4363

4364

307721

static Unidata_decomp *get_decomposition(my_wc_t ch) {

4365

3883031

auto comp_func = [](Unidata_decomp x, Unidata_decomp y) {

4366

3883031

return x.charcode < y.charcode;

4367

};

4368

307721

Unidata_decomp to_find = {ch, CHAR_CATEGORY_LU, DECOMP_TAG_NONE, {0}};

4369

1/2

✓ Branch 0 taken 307721 times.

✗ Branch 1 not taken.

307721

Unidata_decomp *decomp = std::lower_bound(

4370

std::begin(uni_dec), std::end(uni_dec), to_find, comp_func);

4371

5/6

✓ Branch 0 taken 307721 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 49383 times.

✓ Branch 3 taken 258338 times.

✓ Branch 4 taken 49383 times.

✓ Branch 5 taken 258338 times.

307721

if (decomp == std::end(uni_dec) || decomp->charcode != ch) return nullptr;

4372

258338

return decomp;

4373

}

4374

4375

3645724

static Combining_mark *my_find_combining_mark(my_wc_t code) {

4376

35679820

auto comp_func = [](Combining_mark x, Combining_mark y) {

4377

35679820

return x.charcode < y.charcode;

4378

};

4379

3645724

Combining_mark to_find = {code, 0};

4380

1/2

✓ Branch 0 taken 3645724 times.

✗ Branch 1 not taken.

3645724

return std::lower_bound(std::begin(combining_marks),

4381

7291448

std::end(combining_marks), to_find, comp_func);

}

/**

Check if a list of combining marks contains the whole list of origin

4386

decomposed combining marks.

4387

@param origin_dec The origin list of combining marks decomposed from

4388

character in tailoring rule.

4389

@param dec_codes The list of combining marks decomposed from

4390

character in decomposition list.

4391

@param dec_diff The combining marks exist in dec_codes but not in

4392

origin_dec.

4393

@return Whether the list of combining marks contains the

4394

whole list of origin combining marks.

4395

*/

4396

284165408

static bool my_is_inheritance_of_origin(const my_wc_t *origin_dec,

4397

const my_wc_t *dec_codes,

4398

my_wc_t *dec_diff) {

4399

int ind0, ind1, ind2;

4400

2/2

✓ Branch 0 taken 282344219 times.

✓ Branch 1 taken 1821189 times.

284165408

if (origin_dec[0] != dec_codes[0]) return false;

4401

1/2

✓ Branch 0 taken 2710035 times.

✗ Branch 1 not taken.

4531224

for (ind0 = ind1 = ind2 = 1; ind0 < MY_UCA_MAX_CONTRACTION &&

4402

2710035

ind1 < MY_UCA_MAX_CONTRACTION &&

4403

5/6

✓ Branch 0 taken 2710035 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 2456908 times.

✓ Branch 3 taken 253127 times.

✓ Branch 4 taken 2005938 times.

✓ Branch 5 taken 450970 times.

5420070

origin_dec[ind0] && dec_codes[ind1];) {

4404

2/2

✓ Branch 0 taken 183076 times.

✓ Branch 1 taken 1822862 times.

2005938

if (origin_dec[ind0] == dec_codes[ind1]) {

4405

183076

ind0++;

4406

183076

ind1++;

4407

} else {

4408

1822862

Combining_mark *mark0 = my_find_combining_mark(origin_dec[ind0]);

4409

1822862

Combining_mark *mark1 = my_find_combining_mark(dec_codes[ind1]);

4410

2/2

✓ Branch 0 taken 1117092 times.

✓ Branch 1 taken 705770 times.

1822862

if (mark0->ccc == mark1->ccc) return false;

4411

705770

dec_diff[ind2++] = dec_codes[ind1++];

4412

}

4413

}

4414

3/4

✓ Branch 0 taken 704097 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 253127 times.

✓ Branch 3 taken 450970 times.

704097

if (ind0 >= MY_UCA_MAX_CONTRACTION || !origin_dec[ind0]) {

4415

2/2

✓ Branch 0 taken 1075303 times.

✓ Branch 1 taken 253127 times.

1328430

while (ind1 < MY_UCA_MAX_CONTRACTION) {

4416

1075303

dec_diff[ind2++] = dec_codes[ind1++];

4417

}

4418

253127

return true;

4419

}

4420

450970

return false;

}

/**

Add new rules recersively if one rule's characters are in decomposition

4425

list.

4426

@param rules The rule list

4427

@param r The rule to check

4428

@param decomp_rec The decomposition of the character in rule.

4429

@param comp_added Bitset which marks whether the comp

4430

character has been added to rule list.

4431

@return 1 Error adding new rules

4432

0 Add rules successfully

4433

*/

4434

307721

static int my_coll_add_inherit_rules(

4435

MY_COLL_RULES *rules, MY_COLL_RULE *r, const Unidata_decomp *decomp_rec,

4436

std::bitset<array_elements(uni_dec)> *comp_added) {

4437

2/2

✓ Branch 0 taken 1760779562 times.

✓ Branch 1 taken 307721 times.

1761087283

for (uint dec_ind = 0; dec_ind < array_elements(uni_dec); dec_ind++) {

4438

/*

4439

For normal character which can be decomposed, it is always decomposed to

4440

be another character and one combining mark.

4441

4442

Currently we only support the weight inheritance of character that can be

4443

canonical-decomposed to another character and a list of combining marks.

4444

So skip the compatibility decomposition.

4445

4446

Sample from UnicodeData.txt:

4447

Canonical decomposition: U+00DC : U+0055 U+0308

4448

Compatibility decompsition: U+FF59 : <wide> U+0079

4449

*/

4450

9/10

✓ Branch 0 taken 633905260 times.

✓ Branch 1 taken 1126874302 times.

✓ Branch 2 taken 633905260 times.

✗ Branch 3 not taken.

✓ Branch 4 taken 624302842 times.

✓ Branch 5 taken 9602418 times.

✓ Branch 6 taken 523209339 times.

✓ Branch 7 taken 101093503 times.

✓ Branch 8 taken 1476614154 times.

✓ Branch 9 taken 284165408 times.

2283988901

if (!my_compchar_is_normal_char(dec_ind) || comp_added->test(dec_ind) ||

4451

523209339

(decomp_rec != nullptr &&

4452

2/2

✓ Branch 0 taken 340137434 times.

✓ Branch 1 taken 183071905 times.

523209339

uni_dec[dec_ind].decomp_tag != decomp_rec->decomp_tag))

4453

1476614154

continue;

4454

/*

4455

In DUCET, all accented character's weight is defined as base

4456

character's weight followed by accent mark's weight. For example:

4457

00DC = 0055 + 0308

4458

0055 ; [.1E30.0020.0008] # LATIN CAPITAL LETTER U

4459

0308 ; [.0000.002B.0002] # COMBINING DIAERESIS

4460

00DC ; [.1E30.0020.0008][.0000.002B.0002] # LATIN CAPITAL LETTER U

4461

WITH DIAERESIS

4462

So the composition character's rule should be same as origin rule

4463

except of the change of curr value.

4464

*/

4465

284165408

my_wc_t dec_diff[MY_UCA_MAX_CONTRACTION]{r->curr[0], 0};

4466

284165408

my_wc_t orig_dec[MY_UCA_MAX_CONTRACTION]{0};

4467

2/2

✓ Branch 0 taken 101093503 times.

✓ Branch 1 taken 183071905 times.

284165408

if (decomp_rec == nullptr) {

4468

/*

4469

If there is no decomposition record found in Unidata_decomp, it means

4470

its decomposition form is itself.

4471

*/

4472

101093503

orig_dec[0] = r->curr[0];

4473

} else {

4474

183071905

memcpy(orig_dec, decomp_rec->dec_codes, sizeof(orig_dec));

4475

}

4476

1/2

✓ Branch 0 taken 284165408 times.

✗ Branch 1 not taken.

284165408

if (my_is_inheritance_of_origin(orig_dec, uni_dec[dec_ind].dec_codes,

4477

4/4

✓ Branch 0 taken 253127 times.

✓ Branch 1 taken 283912281 times.

✓ Branch 2 taken 154431 times.

✓ Branch 3 taken 284010977 times.

284418535

dec_diff) &&

4478

2/2

✓ Branch 0 taken 154431 times.

✓ Branch 1 taken 98696 times.

253127

!my_comp_in_rulelist(rules, uni_dec[dec_ind].charcode)) {

4479

154431

MY_COLL_RULE newrule{{0}, {uni_dec[dec_ind].charcode, 0}, {0}, 0, false};

4480

154431

memcpy(newrule.base, dec_diff, sizeof(newrule.base));

4481

2/4

✓ Branch 0 taken 154431 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 154431 times.

154431

if (my_coll_rules_add(rules, &newrule)) return 1;

4482

1/2

✓ Branch 0 taken 154431 times.

✗ Branch 1 not taken.

154431

comp_added->set(dec_ind);

4483

}

4484

}

4485

307721

return 0;

4486

}

4487

4488

313946

static bool combining_mark_in_rulelist(const my_wc_t *dec_codes,

4489

const MY_COLL_RULE *r_start,

4490

const MY_COLL_RULE *r_end) {

4491

1/2

✓ Branch 0 taken 466416 times.

✗ Branch 1 not taken.

466416

for (int i = 1; i < MY_UCA_MAX_CONTRACTION; ++i) {

4492

2/2

✓ Branch 0 taken 279818 times.

✓ Branch 1 taken 186598 times.

466416

if (!*(dec_codes + i)) return false;

4493

2/2

✓ Branch 0 taken 12146882 times.

✓ Branch 1 taken 152470 times.

12299352

for (const MY_COLL_RULE *r = r_start; r < r_end; ++r) {

4494

2/2

✓ Branch 0 taken 34128 times.

✓ Branch 1 taken 12112754 times.

12146882

if (r->curr[0] == *(dec_codes + i)) {

4495

34128

return true;

}

}

}

✗

return false;

4500

}

4501

4502

3023

static int add_normalization_rules(const CHARSET_INFO *cs,

4503

MY_COLL_RULES *rules) {

4504

3/4

✓ Branch 0 taken 3023 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 2865 times.

✓ Branch 3 taken 158 times.

3023

if (!cs->coll_param || !cs->coll_param->norm_enabled) return 0;

4505

158

const int orig_rule_num = rules->nrules;

4506

2/2

✓ Branch 0 taken 904076 times.

✓ Branch 1 taken 158 times.

904234

for (Unidata_decomp *decomp = std::begin(uni_dec); decomp < std::end(uni_dec);

4507

++decomp) {

4508

2/2

✓ Branch 0 taken 313946 times.

✓ Branch 1 taken 11534 times.

1229556

if (!my_compchar_is_normal_char(decomp) ||

4509

4/4

✓ Branch 0 taken 325480 times.

✓ Branch 1 taken 578596 times.

✓ Branch 2 taken 869948 times.

✓ Branch 3 taken 34128 times.

1229556

my_comp_in_rulelist(rules, decomp->charcode) ||

4510

2/2

✓ Branch 0 taken 279818 times.

✓ Branch 1 taken 34128 times.

313946

!combining_mark_in_rulelist(decomp->dec_codes, rules->rule,

4511

313946

rules->rule + orig_rule_num))

4512

869948

continue;

4513

34128

MY_COLL_RULE newrule{{0}, {decomp->charcode, 0}, {0}, 0, false};

4514

34128

memcpy(newrule.base, decomp->dec_codes, sizeof(newrule.base));

4515

2/4

✓ Branch 0 taken 34128 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 34128 times.

34128

if (my_coll_rules_add(rules, &newrule)) return 1;

4516

}

4517

158

return 0;

}

/**

For every rule in rule list, check and add new rules if it is in

4522

decomposition list.

4523

@param cs Character set info

4524

@param rules The rule list

4525

@return 1 Error happens when adding new rule

4526

0 Add rules successfully

4527

*/

4528

29006

static int my_coll_check_rule_and_inherit(const CHARSET_INFO *cs,

4529

MY_COLL_RULES *rules) {

4530

2/2

✓ Branch 0 taken 19647 times.

✓ Branch 1 taken 9359 times.

29006

if (rules->uca->version != UCA_V900) return 0;

4531

4532

/*

4533

Character can combine with marks to be a new character. For example,

4534

A + [mark b] = A1, A1 + [mark c] = A2. We think the weight of A1 and

4535

A2 should shift with A if A is in rule list and its weight shifts,

4536

unless A1 / A2 is already in rule list.

4537

*/

4538

9359

std::bitset<array_elements(uni_dec)> comp_added;

4539

9359

int orig_rule_num = rules->nrules;

4540

2/2

✓ Branch 0 taken 546208 times.

✓ Branch 1 taken 9359 times.

555567

for (int i = 0; i < orig_rule_num; ++i) {

4541

546208

MY_COLL_RULE r = *(rules->rule + i);

4542

/*

4543

Do not add inheritance rule for contraction.

4544

But for the Chinese collation, the weight shift rule of Chinese collation

4545

is a bit different from all the languages we added so far. For example, it

4546

has a rule "&e << ... << e\\u0302\\u0300". So far, if a language's rule

4547

involves 'e\\u0302\\u0300', it will use the combining form character,

4548

U+1EC1, and it is not a contraction. If we don't handle this for Chinese

4549

collation, it will skip some further rule inheriting.

4550

*/

4551

4/4

✓ Branch 0 taken 462048 times.

✓ Branch 1 taken 84160 times.

✓ Branch 2 taken 238487 times.

✓ Branch 3 taken 223561 times.

546208

if (cs->coll_param != &zh_coll_param && r.curr[1]) continue;

4552

1/2

✓ Branch 0 taken 307721 times.

✗ Branch 1 not taken.

307721

Unidata_decomp *decomp_rec = get_decomposition(r.curr[0]);

4553

2/4

✓ Branch 0 taken 307721 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 307721 times.

307721

if (my_coll_add_inherit_rules(rules, &r, decomp_rec, &comp_added)) return 1;

4554

}

4555

9359

return 0;

}

/**

Helper function to store weight boundary values.

4560

@param[out] wt_rec Weight boundary for each character group and gap

4561

between groups

4562

@param rec_ind The position from where to store weight boundary

4563

@param old_begin Beginning weight of character group before reorder

4564

@param old_end End weight of character group before reorder

4565

@param new_begin Beginning weight of character group after reorder

4566

@param new_end End weight of character group after reorder

4567

*/

4568

5735

static inline void my_set_weight_rec(

4569

Reorder_wt_rec (&wt_rec)[2 * UCA_MAX_CHAR_GRP], int rec_ind,

4570

uint16 old_begin, uint16 old_end, uint16 new_begin, uint16 new_end) {

4571

5735

wt_rec[rec_ind] = {{old_begin, old_end}, {new_begin, new_end}};

4572

5735

}

4573

4574

/**

4575

Calculate the reorder parameters for the character groups.

4576

@param cs Character set info

4577

@param[out] rec_ind The position from where to store weight boundary

4578

*/

4579

2230

static void my_calc_char_grp_param(const CHARSET_INFO *cs, int &rec_ind) {

4580

2230

int weight_start = START_WEIGHT_TO_REORDER;

4581

2230

int grp_ind = 0;

4582

2230

Reorder_param *param = cs->coll_param->reorder_param;

4583

1/2

✓ Branch 0 taken 5735 times.

✗ Branch 1 not taken.

5735

for (; grp_ind < UCA_MAX_CHAR_GRP; ++grp_ind) {

4584

2/2

✓ Branch 0 taken 2230 times.

✓ Branch 1 taken 3505 times.

5735

if (param->reorder_grp[grp_ind] == CHARGRP_NONE) break;

4585

9886

for (Char_grp_info *info = std::begin(char_grp_infos);

4586

1/2

✓ Branch 0 taken 9886 times.

✗ Branch 1 not taken.

9886

info < std::end(char_grp_infos); ++info) {

4587

2/2

✓ Branch 0 taken 6381 times.

✓ Branch 1 taken 3505 times.

9886

if (param->reorder_grp[grp_ind] != info->group) continue;

4588

3505

my_set_weight_rec(

4589

3505

param->wt_rec, grp_ind, info->grp_wt_bdy.begin, info->grp_wt_bdy.end,

4590

weight_start,

4591

3505

weight_start + info->grp_wt_bdy.end - info->grp_wt_bdy.begin);

4592

3505

weight_start = param->wt_rec[grp_ind].new_wt_bdy.end + 1;

4593

3505

break;

4594

}

4595

}

4596

2230

rec_ind = grp_ind;

4597

2230

}

4598

4599

/**

4600

Calculate the reorder parameters for the gap between character groups.

4601

@param cs Character set info

4602

@param rec_ind The position from where to store weight boundary

4603

*/

4604

2230

static void my_calc_char_grp_gap_param(CHARSET_INFO *cs, int &rec_ind) {

4605

2230

Reorder_param *param = cs->coll_param->reorder_param;

4606

2230

uint16 weight_start = param->wt_rec[rec_ind - 1].new_wt_bdy.end + 1;

4607

2230

Char_grp_info *last_grp = nullptr;

4608

13380

for (Char_grp_info *info = std::begin(char_grp_infos);

4609

2/2

✓ Branch 0 taken 11150 times.

✓ Branch 1 taken 2230 times.

13380

info < std::end(char_grp_infos); ++info) {

4610

1/2

✓ Branch 0 taken 23895 times.

✗ Branch 1 not taken.

23895

for (int ind = 0; ind < UCA_MAX_CHAR_GRP; ++ind) {

4611

2/2

✓ Branch 0 taken 7645 times.

✓ Branch 1 taken 16250 times.

23895

if (param->reorder_grp[ind] == CHARGRP_NONE) break;

4612

2/2

✓ Branch 0 taken 12745 times.

✓ Branch 1 taken 3505 times.

16250

if (param->reorder_grp[ind] != info->group) continue;

4613

2/2

✓ Branch 0 taken 794 times.

✓ Branch 1 taken 2711 times.

3505

if (param->max_weight < info->grp_wt_bdy.end)

4614

794

param->max_weight = info->grp_wt_bdy.end;

4615

/*

4616

There might be some character groups before the first character

4617

group in our list.

4618

*/

4619

4/4

✓ Branch 0 taken 2230 times.

✓ Branch 1 taken 1275 times.

✓ Branch 2 taken 955 times.

✓ Branch 3 taken 1275 times.

3505

if (!last_grp && info->grp_wt_bdy.begin > START_WEIGHT_TO_REORDER) {

4620

955

my_set_weight_rec(param->wt_rec, rec_ind, START_WEIGHT_TO_REORDER,

4621

955

info->grp_wt_bdy.begin - 1, weight_start,

4622

955

weight_start + (info->grp_wt_bdy.begin - 1) -

4623

START_WEIGHT_TO_REORDER);

4624

955

weight_start = param->wt_rec[rec_ind].new_wt_bdy.end + 1;

4625

955

rec_ind++;

4626

}

4627

/* Gap between 2 character groups in out list. */

4628

3/4

✓ Branch 0 taken 1275 times.

✓ Branch 1 taken 2230 times.

✓ Branch 2 taken 1275 times.

✗ Branch 3 not taken.

3505

if (last_grp && last_grp->grp_wt_bdy.end < (info->grp_wt_bdy.begin - 1)) {

4629

1275

my_set_weight_rec(param->wt_rec, rec_ind, last_grp->grp_wt_bdy.end + 1,

4630

1275

info->grp_wt_bdy.begin - 1, weight_start,

4631

1275

weight_start + (info->grp_wt_bdy.begin - 1) -

4632

1275

(last_grp->grp_wt_bdy.end + 1));

4633

1275

weight_start = param->wt_rec[rec_ind].new_wt_bdy.end + 1;

4634

1275

rec_ind++;

4635

}

4636

3505

last_grp = info;

4637

3505

break;

4638

}

4639

}

4640

2230

param->wt_rec_num = rec_ind;

4641

2230

}

4642

4643

/**

4644

Prepare reorder parameters.

4645

@param cs Character set info

4646

*/

4647

3023

static int my_prepare_reorder(CHARSET_INFO *cs) {

4648

/*

4649

Chinese collation's reordering is done in next_implicit() and

4650

modify_all_zh_pages(). See the comment on zh_reorder_param and

4651

change_zh_implicit().

4652

*/

4653

4/4

✓ Branch 0 taken 2390 times.

✓ Branch 1 taken 633 times.

✓ Branch 2 taken 160 times.

✓ Branch 3 taken 2230 times.

3023

if (!cs->coll_param->reorder_param || cs->coll_param == &zh_coll_param)

4654

793

return 0;

4655

/*

4656

For each group of character, for example, latin characters,

4657

their weights are in a separate range. The default sequence

4658

of these groups is: Latin, Greek, Coptic, Cyrillic, and so

4659

on. Some languages want to change the default sequence. For

4660

example, Croatian wants to put Cyrillic to just behind Latin.

4661

We need to reorder the character groups and change their

4662

weight accordingly. Here we calculate the parameters needed

4663

for weight change. And the change will happen when weight

4664

returns from strnxfrm.

4665

*/

4666

2230

int rec_ind = 0;

4667

2230

my_calc_char_grp_param(cs, rec_ind);

4668

2230

my_calc_char_grp_gap_param(cs, rec_ind);

4669

2230

return rec_ind;

4670

}

4671

4672

323

static void adjust_japanese_weight(CHARSET_INFO *cs, int rec_ind) {

4673

/*

4674

Per CLDR 30, Japanese collations need to reorder characters as

4675

[Latin, Kana, Han, others]. So for the original character group list:

4676

[Latin, CharA, Kana, CharB, Han, Others], it should be reordered as

4677

[Latin, Kana, Han, CharA, CharB, Others]. But my_prepare_reorder()

4678

reorders original group to be [Latin, Kana, CharA, CharB, Han, Others].

4679

This is because Han characters are different from others in that Han

4680

characters' weight is implicit and has two primary weights for each

4681

character. Other characters have only one primary weight for each (base)

4682

character. Han characters always sort bigger.

4683

4684

CLDR defines the collating order for 6355 Japanese Han characters. All

4685

of them are in [U+4E00, U+9FFF]; we give them tailored primary weights

4686

in ja_han_pages. The tailored primary weights are just after Kana,

4687

because these characters are very common. These Han characters' weight

4688

pages will be added to collation's UCA data in copy_ja_han_pages().

4689

For the other Han characters, we don't change their implicit weights,

4690

which is [FB80 - FB85, 0020, 0002][XXXX, 0000, 0000].

4691

4692

To make sure CharA and CharB's weight is greater than all Han characters,

4693

we give them weight as [FB86, 0000, 0000][origin weights]. This will be

4694

done in apply_reorder_param().

4695

4696

Because the values stored in last wt_rec element is calculated for moving

4697

CharA to be after Kana, but we want them to be after all Han character,

4698

we reset the weight boundary here, and will change all these characters'

4699

weight in apply_reorder_param().

4700

*/

4701

323

Reorder_param *param = cs->coll_param->reorder_param;

4702

323

param->wt_rec[rec_ind - 1].new_wt_bdy.begin = 0;

4703

323

param->wt_rec[rec_ind - 1].new_wt_bdy.end = 0;

4704

323

param->wt_rec[rec_ind].old_wt_bdy.begin = param->wt_rec[1].old_wt_bdy.end + 1;

4705

323

param->wt_rec[rec_ind].old_wt_bdy.end = 0x54A3;

4706

323

param->wt_rec[rec_ind].new_wt_bdy.begin = 0;

4707

323

param->wt_rec[rec_ind].new_wt_bdy.end = 0;

4708

323

param->wt_rec_num++;

4709

323

param->max_weight = 0x54A3;

4710

323

}

4711

4712

/**

4713

Prepare parametric tailoring, like reorder, etc.

4714

@param cs Character set info

4715

@param rules Collation rule list to add to.

4716

@return false Collation parameters applied successfully.

4717

true Error happened.

4718

*/

4719

29006

static bool my_prepare_coll_param(CHARSET_INFO *cs, MY_COLL_RULES *rules) {

4720

4/4

✓ Branch 0 taken 9359 times.

✓ Branch 1 taken 19647 times.

✓ Branch 2 taken 6336 times.

✓ Branch 3 taken 3023 times.

29006

if (rules->uca->version != UCA_V900 || !cs->coll_param) return false;

4721

4722

3023

int rec_ind = my_prepare_reorder(cs);

4723

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 3023 times.

3023

if (add_normalization_rules(cs, rules)) return true;

4724

4725

2/2

✓ Branch 0 taken 323 times.

✓ Branch 1 taken 2700 times.

3023

if (cs->coll_param == &ja_coll_param) adjust_japanese_weight(cs, rec_ind);

4726

/* Might add other parametric tailoring rules later. */

4727

3023

return false;

}

/*

This function copies an UCS2 collation from

4732

the default Unicode Collation Algorithm (UCA)

4733

weights applying tailorings, i.e. a set of

4734

alternative weights for some characters.

4735

4736

The default UCA weights are stored in uca_weight/uca_length.

4737

They consist of 256 pages, 256 character each.

4738

4739

If a page is not overwritten by tailoring rules,

4740

it is copies as is from UCA as is.

4741

4742

If a page contains some overwritten characters, it is

4743

allocated. Untouched characters are copied from the

default weights.

*/

136555

static bool create_tailoring(CHARSET_INFO *cs, MY_CHARSET_LOADER *loader) {

4748

2/2

✓ Branch 0 taken 107546 times.

✓ Branch 1 taken 29009 times.

136555

if (!cs->tailoring)

4749

107546

return false; /* Ok to add a collation without tailoring */

4750

4751

MY_COLL_RULES rules;

4752

29009

MY_UCA_INFO new_uca, *src_uca = nullptr;

4753

29009

int rc = 0;

4754

MY_UCA_INFO *src, *dst;

4755

size_t npages;

4756

bool lengths_are_temporary;

4757

4758

29009

loader->errcode = 0;

4759

29009

*loader->errarg = '\0';

4760

4761

29009

memset(&rules, 0, sizeof(rules));

4762

29009

rules.loader = loader;

4763

1/2

✓ Branch 0 taken 29009 times.

✗ Branch 1 not taken.

29009

rules.uca = cs->uca ? cs->uca : &my_uca_v400; /* For logical positions, etc */

4764

29009

memset(&new_uca, 0, sizeof(new_uca));

4765

4766

/* Parse ICU Collation Customization expression */

4767

2/2

✓ Branch 0 taken 3 times.

✓ Branch 1 taken 29006 times.

29009

if ((rc = my_coll_rule_parse(&rules, cs->tailoring,

4768

1/2

✓ Branch 0 taken 29009 times.

✗ Branch 1 not taken.

29009

cs->tailoring + strlen(cs->tailoring),

4769

cs->m_coll_name)))

4770

3

goto ex;

4771

4772

2/4

✓ Branch 0 taken 29006 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 29006 times.

29006

if ((rc = my_coll_check_rule_and_inherit(cs, &rules))) goto ex;

4773

4774

2/4

✓ Branch 0 taken 29006 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 29006 times.

29006

if ((rc = my_prepare_coll_param(cs, &rules))) goto ex;

4775

4776

2/2

✓ Branch 0 taken 972 times.

✓ Branch 1 taken 28034 times.

29006

if (rules.uca->version == UCA_V520) /* Unicode-5.2.0 requested */

4777

{

4778

972

src_uca = &my_uca_v520;

4779

972

cs->caseinfo = &my_unicase_unicode520;

4780

2/2

✓ Branch 0 taken 18675 times.

✓ Branch 1 taken 9359 times.

28034

} else if (rules.uca->version == UCA_V400) /* Unicode-4.0.0 requested */

4781

{

4782

18675

src_uca = &my_uca_v400;

4783

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 18675 times.

18675

if (!cs->caseinfo) cs->caseinfo = &my_unicase_default;

4784

} else /* No Unicode version specified */

4785

{

4786

1/2

✓ Branch 0 taken 9359 times.

✗ Branch 1 not taken.

9359

src_uca = cs->uca ? cs->uca : &my_uca_v400;

4787

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 9359 times.

9359

if (!cs->caseinfo) cs->caseinfo = &my_unicase_default;

}

/*

For UCA 9.0.0, we don't have a length page, but we still create one

4792

temporarily so that we can keep track of much memory we need to

4793

allocate for weights.

4794

*/

4795

29006

src = src_uca;

4796

29006

dst = &new_uca;

4797

4798

29006

dst->extra_ce_pri_base = cs->uca->extra_ce_pri_base;

4799

29006

dst->extra_ce_sec_base = cs->uca->extra_ce_sec_base;

4800

29006

dst->extra_ce_ter_base = cs->uca->extra_ce_ter_base;

4801

4/4

✓ Branch 0 taken 3023 times.

✓ Branch 1 taken 25983 times.

✓ Branch 2 taken 160 times.

✓ Branch 3 taken 2863 times.

29006

if (cs->coll_param && cs->coll_param == &zh_coll_param) {

4802

160

dst->extra_ce_pri_base = ZH_EXTRA_CE_PRI;

4803

}

4804

4805

29006

npages = (src->maxchar + 1) / 256;

4806

2/2

✓ Branch 0 taken 9359 times.

✓ Branch 1 taken 19647 times.

29006

if (rules.uca->version == UCA_V900) {

4807

2/4

✓ Branch 0 taken 9359 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 9359 times.

9359

if (!(src->lengths = (uchar *)(loader->mem_malloc)(npages))) goto ex;

4808

1/2

✓ Branch 0 taken 9359 times.

✗ Branch 1 not taken.

9359

synthesize_lengths_900(src->lengths, src->weights, npages);

4809

}

4810

4811

29006

lengths_are_temporary = (rules.uca->version == UCA_V900);

4812

3/4

✓ Branch 0 taken 29006 times.

✗ Branch 1 not taken.

✓ Branch 2 taken 2 times.

✓ Branch 3 taken 29004 times.

29006

if ((rc = init_weight_level(cs, loader, &rules, 0, dst, src,

4813

lengths_are_temporary)))

4814

2

goto ex;

4815

4816

2/2

✓ Branch 0 taken 9359 times.

✓ Branch 1 taken 19645 times.

29004

if (lengths_are_temporary) {

4817

1/2

✓ Branch 0 taken 9359 times.

✗ Branch 1 not taken.

9359

(loader->mem_free)(src->lengths);

4818

1/2

✓ Branch 0 taken 9359 times.

✗ Branch 1 not taken.

9359

(loader->mem_free)(dst->lengths);

4819

9359

src->lengths = nullptr;

4820

9359

dst->lengths = nullptr;

4821

}

4822

4823

29004

new_uca.version = src_uca->version;

4824

2/4

✓ Branch 0 taken 29004 times.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 29004 times.

29004

if (!(cs->uca = (MY_UCA_INFO *)(loader->once_alloc)(sizeof(MY_UCA_INFO)))) {

4825

✗

rc = 1;

4826

✗

goto ex;

4827

}

4828

29004

memset(cs->uca, 0, sizeof(MY_UCA_INFO));

4829

29004

cs->uca[0] = new_uca;

4830

4831

29009

ex:

4832

1/2

✓ Branch 0 taken 29009 times.

✗ Branch 1 not taken.

29009

(loader->mem_free)(rules.rule);

4833

3/4

✓ Branch 0 taken 5 times.

✓ Branch 1 taken 29004 times.

✓ Branch 2 taken 5 times.

✗ Branch 3 not taken.

29009

if (rc != 0 && loader->errcode) {

4834

3/4

✓ Branch 0 taken 2 times.

✓ Branch 1 taken 3 times.

✓ Branch 2 taken 2 times.

✗ Branch 3 not taken.

5

if (new_uca.contraction_nodes) delete new_uca.contraction_nodes;

4835

1/2

✓ Branch 0 taken 5 times.

✗ Branch 1 not taken.

5

loader->reporter(ERROR_LEVEL, loader->errcode, loader->errarg);

4836

}

4837

29009

return rc;

4838

}

4839

4840

19367193

static void my_coll_uninit_uca(CHARSET_INFO *cs) {

4841

4/4

✓ Branch 0 taken 7215187 times.

✓ Branch 1 taken 12152006 times.

✓ Branch 2 taken 8842 times.

✓ Branch 3 taken 7206345 times.

19367193

if (cs->uca && cs->uca->contraction_nodes) {

4842

1/2

✓ Branch 0 taken 8842 times.

✗ Branch 1 not taken.

8842

delete cs->uca->contraction_nodes;

4843

8842

cs->uca->contraction_nodes = nullptr;

4844

8842

cs->state &= ~MY_CS_READY;

4845

}

4846

19367193

}

4847

/*

4848

Universal CHARSET_INFO compatible wrappers

4849

for the above internal functions.

4850

Should work for any character set.

*/

extern "C" {

136555

static bool my_coll_init_uca(CHARSET_INFO *cs, MY_CHARSET_LOADER *loader) {

4855

136555

cs->pad_char = ' ';

4856

136555

cs->ctype = my_charset_utf8_unicode_ci.ctype;

4857

2/2

✓ Branch 0 taken 39 times.

✓ Branch 1 taken 136516 times.

136555

if (!cs->caseinfo) cs->caseinfo = &my_unicase_default;

4858

2/2

✓ Branch 0 taken 18680 times.

✓ Branch 1 taken 117875 times.

136555

if (!cs->uca) cs->uca = &my_uca_v400;

4859

136555

return create_tailoring(cs, loader);

4860

}

4861

4862

80

static int my_strnncoll_any_uca(const CHARSET_INFO *cs, const uchar *s,

4863

size_t slen, const uchar *t, size_t tlen,

4864

bool t_is_prefix) {

4865

2/2

✓ Branch 0 taken 54 times.

✓ Branch 1 taken 26 times.

80

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

4866

1/2

✓ Branch 0 taken 54 times.

✗ Branch 1 not taken.

54

return my_strnncoll_uca<uca_scanner_any<Mb_wc_utf8mb4>, 1>(

4867

54

cs, Mb_wc_utf8mb4(), s, slen, t, tlen, t_is_prefix);

4868

}

4869

4870

26

Mb_wc_through_function_pointer mb_wc(cs);

4871

1/2

✓ Branch 0 taken 26 times.

✗ Branch 1 not taken.

26

return my_strnncoll_uca<uca_scanner_any<decltype(mb_wc)>, 1>(

4872

26

cs, mb_wc, s, slen, t, tlen, t_is_prefix);

4873

}

4874

4875

264540

static int my_strnncollsp_any_uca(const CHARSET_INFO *cs, const uchar *s,

4876

size_t slen, const uchar *t, size_t tlen) {

4877

2/2

✓ Branch 0 taken 145742 times.

✓ Branch 1 taken 118798 times.

264540

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

4878

1/2

✓ Branch 0 taken 145742 times.

✗ Branch 1 not taken.

145742

return my_strnncollsp_uca(cs, Mb_wc_utf8mb4(), s, slen, t, tlen);

4879

}

4880

4881

118798

Mb_wc_through_function_pointer mb_wc(cs);

4882

1/2

✓ Branch 0 taken 118811 times.

✗ Branch 1 not taken.

118796

return my_strnncollsp_uca(cs, mb_wc, s, slen, t, tlen);

4883

}

4884

4885

2324

static void my_hash_sort_any_uca(const CHARSET_INFO *cs, const uchar *s,

4886

size_t slen, uint64 *n1, uint64 *n2) {

4887

2/2

✓ Branch 0 taken 648 times.

✓ Branch 1 taken 1676 times.

2324

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

4888

648

my_hash_sort_uca(cs, Mb_wc_utf8mb4(), s, slen, n1, n2);

4889

} else {

4890

1676

Mb_wc_through_function_pointer mb_wc(cs);

4891

1/2

✓ Branch 0 taken 1676 times.

✗ Branch 1 not taken.

1676

my_hash_sort_uca(cs, mb_wc, s, slen, n1, n2);

4892

}

4893

2324

}

4894

4895

248749863

static size_t my_strnxfrm_any_uca(const CHARSET_INFO *cs, uchar *dst,

4896

size_t dstlen, uint num_codepoints,

4897

const uchar *src, size_t srclen, uint flags) {

4898

2/2

✓ Branch 0 taken 80229934 times.

✓ Branch 1 taken 168519929 times.

248749863

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

4899

1/2

✓ Branch 0 taken 80229934 times.

✗ Branch 1 not taken.

80229934

return my_strnxfrm_uca(cs, Mb_wc_utf8mb4(), dst, dstlen, num_codepoints,

4900

80229934

src, srclen, flags);

4901

}

4902

4903

168519929

Mb_wc_through_function_pointer mb_wc(cs);

4904

1/2

✓ Branch 0 taken 168519929 times.

✗ Branch 1 not taken.

168519929

return my_strnxfrm_uca(cs, mb_wc, dst, dstlen, num_codepoints, src, srclen,

4905

168519929

flags);

4906

}

4907

4908

5211575177

static int my_strnncoll_uca_900(const CHARSET_INFO *cs, const uchar *s,

4909

size_t slen, const uchar *t, size_t tlen,

4910

bool t_is_prefix) {

4911

1/2

✓ Branch 0 taken 5211600671 times.

✗ Branch 1 not taken.

5211575177

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

4912

3/5

✓ Branch 0 taken 5211594076 times.

✓ Branch 1 taken 4584 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 2944 times.

✗ Branch 4 not taken.

5211600671

switch (cs->levels_for_compare) {

4913

5211594076

case 1:

4914

1/2

✓ Branch 0 taken 5211879234 times.

✗ Branch 1 not taken.

5211594076

return my_strnncoll_uca<uca_scanner_900<Mb_wc_utf8mb4, 1>, 1>(

4915

5211879234

cs, Mb_wc_utf8mb4(), s, slen, t, tlen, t_is_prefix);

4916

4584

case 2:

4917

1/2

✓ Branch 0 taken 4584 times.

✗ Branch 1 not taken.

4584

return my_strnncoll_uca<uca_scanner_900<Mb_wc_utf8mb4, 2>, 2>(

4918

4584

cs, Mb_wc_utf8mb4(), s, slen, t, tlen, t_is_prefix);

4919

✗

default:

4920

✗

assert(false);

4921

2944

case 3:

4922

1/2

✓ Branch 0 taken 2944 times.

✗ Branch 1 not taken.

2944

return my_strnncoll_uca<uca_scanner_900<Mb_wc_utf8mb4, 3>, 3>(

4923

2944

cs, Mb_wc_utf8mb4(), s, slen, t, tlen, t_is_prefix);

4924

745

case 4:

4925

1/2

✓ Branch 0 taken 62 times.

✗ Branch 1 not taken.

745

return my_strnncoll_uca<uca_scanner_900<Mb_wc_utf8mb4, 4>, 4>(

4926

62

cs, Mb_wc_utf8mb4(), s, slen, t, tlen, t_is_prefix);

}

}

✗

Mb_wc_through_function_pointer mb_wc(cs);

4931

✗

switch (cs->levels_for_compare) {

4932

✗

case 1:

4933

✗

return my_strnncoll_uca<uca_scanner_900<decltype(mb_wc), 1>, 1>(

4934

✗

cs, mb_wc, s, slen, t, tlen, t_is_prefix);

4935

✗

case 2:

4936

✗

return my_strnncoll_uca<uca_scanner_900<decltype(mb_wc), 2>, 2>(

4937

✗

cs, mb_wc, s, slen, t, tlen, t_is_prefix);

4938

✗

default:

4939

✗

assert(false);

4940

✗

case 3:

4941

✗

return my_strnncoll_uca<uca_scanner_900<decltype(mb_wc), 3>, 3>(

4942

✗

cs, mb_wc, s, slen, t, tlen, t_is_prefix);

4943

✗

case 4:

4944

✗

return my_strnncoll_uca<uca_scanner_900<decltype(mb_wc), 4>, 4>(

4945

✗

cs, mb_wc, s, slen, t, tlen, t_is_prefix);

}

}

812773129

static int my_strnncollsp_uca_900(const CHARSET_INFO *cs, const uchar *s,

4950

size_t slen, const uchar *t, size_t tlen) {

4951

// We are a NO PAD collation, so this is identical to strnncoll.

4952

812773129

return my_strnncoll_uca_900(cs, s, slen, t, tlen, false);

}

} // extern "C"

template <class Mb_wc, int LEVELS_FOR_COMPARE>

4958

303824342

static void my_hash_sort_uca_900_tmpl(const CHARSET_INFO *cs, const Mb_wc mb_wc,

4959

const uchar *s, size_t slen, uint64 *n1) {

4960

303824342

uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE> scanner(mb_wc, cs, s, slen);

4961

4962

/*

4963

A variation of the FNV-1a hash. The differences between this and

4964

standard FNV-1a as described in literature are:

4965

4966

- We work naturally on 16-bit weights, so we XOR in the entire weight

4967

instead of hashing byte-by-byte. (This is effectively a speed/quality

4968

tradeoff, as it will reduce avalanche.)

4969

- We use the n1 seed by XOR-ing it onto the offset basis; FNV-1a as

4970

typically described does not use a seed. This should be safe, since

4971

there's nothing magical about the offset basis; it's just the FNV-1a

4972

hash of some human-readable text.

4973

4974

This is nowhere near a perfect hash function; it has suboptimal avalanche

4975

characteristics, and it not multicollision resistant. In particular,

4976

it fails many SMHasher tests, mostly for bias (collision tests are fine).

4977

However, it is of much better quality than the home-grown hash used

4978

for other collations (which fails _all_ SMHasher tests), while being

4979

much faster.

4980

4981

We ignore the n2 seed entirely, since we don't need it. The caller is

4982

responsible for doing hash folding at the end; we can't do that.

4983

4984

See http://isthe.com/chongo/tech/comp/fnv/#FNV-param for constants.

4985

*/

4986

4987

303824372

uint64 h = *n1;

4988

303824372

h ^= 14695981039346656037ULL;

4989

4990

2/2

✓ Branch 0 taken 151912094 times.

✓ Branch 1 taken 92 times.

303824372

scanner.for_each_weight(

4991

2116889324

[&](int s_res, bool) -> bool {

4992

2116889324

h ^= s_res;

4993

2116889324

h *= 1099511628211ULL;

4994

2116889324

return true;

4995

},

4996

473260726

[](int) { return true; });

4997

4998

303824542

*n1 = h;

}

extern "C" {

151911974

static void my_hash_sort_uca_900(const CHARSET_INFO *cs, const uchar *s,

5004

size_t slen, uint64 *n1, uint64 *) {

5005

1/2

✓ Branch 0 taken 151912101 times.

✗ Branch 1 not taken.

151911974

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

5006

3/5

✓ Branch 0 taken 151912143 times.

✓ Branch 1 taken 14 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 141 times.

✗ Branch 4 not taken.

151912101

switch (cs->levels_for_compare) {

5007

151912143

case 1:

5008

1/2

✓ Branch 0 taken 151912328 times.

✗ Branch 1 not taken.

151912143

return my_hash_sort_uca_900_tmpl<Mb_wc_utf8mb4, 1>(cs, Mb_wc_utf8mb4(),

5009

151912328

s, slen, n1);

5010

14

case 2:

5011

1/2

✓ Branch 0 taken 14 times.

✗ Branch 1 not taken.

14

return my_hash_sort_uca_900_tmpl<Mb_wc_utf8mb4, 2>(cs, Mb_wc_utf8mb4(),

5012

14

s, slen, n1);

5013

✗

default:

5014

✗

assert(false);

5015

141

case 3:

5016

1/2

✓ Branch 0 taken 141 times.

✗ Branch 1 not taken.

141

return my_hash_sort_uca_900_tmpl<Mb_wc_utf8mb4, 3>(cs, Mb_wc_utf8mb4(),

5017

141

s, slen, n1);

5018

✗

case 4:

5019

✗

return my_hash_sort_uca_900_tmpl<Mb_wc_utf8mb4, 4>(cs, Mb_wc_utf8mb4(),

5020

38

s, slen, n1);

}

}

✗

Mb_wc_through_function_pointer mb_wc(cs);

5025

✗

switch (cs->levels_for_compare) {

5026

✗

case 1:

5027

✗

return my_hash_sort_uca_900_tmpl<decltype(mb_wc), 1>(cs, mb_wc, s, slen,

5028

✗

n1);

5029

✗

case 2:

5030

✗

return my_hash_sort_uca_900_tmpl<decltype(mb_wc), 2>(cs, mb_wc, s, slen,

5031

✗

n1);

5032

✗

default:

5033

✗

assert(false);

5034

✗

case 3:

5035

✗

return my_hash_sort_uca_900_tmpl<decltype(mb_wc), 3>(cs, mb_wc, s, slen,

5036

✗

n1);

5037

✗

case 4:

5038

✗

return my_hash_sort_uca_900_tmpl<decltype(mb_wc), 4>(cs, mb_wc, s, slen,

5039

✗

n1);

}

}

} // extern "C"

/*

Check if a constant can be propagated

5047

5048

Currently we don't check the constant itself, and decide not to propagate

5049

a constant just if the collation itself allows expansions or contractions.

5050

*/

5051

189343

bool my_propagate_uca_900(const CHARSET_INFO *cs,

5052

const uchar *str [[maybe_unused]],

5053

size_t length [[maybe_unused]]) {

5054

189343

return !my_uca_have_contractions(cs->uca);

5055

}

5056

5057

template <class Mb_wc, int LEVELS_FOR_COMPARE>

5058

494915754

static size_t my_strnxfrm_uca_900_tmpl(const CHARSET_INFO *cs,

5059

const Mb_wc mb_wc, uchar *dst,

5060

size_t dstlen, const uchar *src,

5061

size_t srclen, uint flags) {

5062

494915754

uchar *d0 = dst;

5063

494915754

uchar *dst_end = dst + dstlen;

5064

494915754

uca_scanner_900<Mb_wc, LEVELS_FOR_COMPARE> scanner(mb_wc, cs, src, srclen);

5065

5066

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 247459484 times.

494916080

assert((dstlen % 2) == 0);

5067

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 247459484 times.

494918968

if ((dstlen % 2) == 1) {

5068

// Emergency workaround for optimized mode.

5069

✗

--dst_end;

5070

}

5071

5072

2/2

✓ Branch 0 taken 247458029 times.

✓ Branch 1 taken 1455 times.

494918968

if (dst != dst_end) {

5073

2/2

✓ Branch 0 taken 50244123 times.

✓ Branch 1 taken 197213906 times.

989826970

scanner.for_each_weight(

5074

6200149933

[&dst, dst_end](int s_res,

5075

bool is_level_separator [[maybe_unused]]) -> bool {

5076

4/16

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✗ Branch 3 not taken.

✗ Branch 4 not taken.

✗ Branch 5 not taken.

✗ Branch 6 not taken.

✗ Branch 7 not taken.

✗ Branch 8 not taken.

✓ Branch 9 taken 20287995 times.

✗ Branch 10 not taken.

✓ Branch 11 taken 624007531 times.

✗ Branch 12 not taken.

✓ Branch 13 taken 13394654 times.

✗ Branch 14 not taken.

✓ Branch 15 taken 1409026496 times.

2066716676

assert(is_level_separator == (s_res == 0));

5077

1/4

✗ Branch 0 not taken.

✗ Branch 1 not taken.

✗ Branch 2 not taken.

✓ Branch 3 taken 1409026496 times.

1409026496

if (LEVELS_FOR_COMPARE == 1) assert(!is_level_separator);

5078

5079

2066716676

dst = store16be(dst, s_res);

5080

2066716581

return (dst < dst_end);

5081

},

5082

313245093

[&dst, dst_end](int num_weights) {

5083

313245093

return (dst < dst_end - num_weights * 2);

});

}

2/2

✓ Branch 0 taken 79838 times.

✓ Branch 1 taken 247377073 times.

494913822

if (flags & MY_STRXFRM_PAD_TO_MAXLEN) {

5088

159676

memset(dst, 0, dst_end - dst);

5089

159676

dst = dst_end;

5090

}

5091

5092

494913822

return dst - d0;

}

extern "C" {

247457290

static size_t my_strnxfrm_uca_900(const CHARSET_INFO *cs, uchar *dst,

5098

size_t dstlen,

5099

uint num_codepoints [[maybe_unused]],

5100

const uchar *src, size_t srclen, uint flags) {

5101

1/2

✓ Branch 0 taken 247457530 times.

✗ Branch 1 not taken.

247457290

if (cs->cset->mb_wc == my_mb_wc_utf8mb4_thunk) {

5102

4/5

✓ Branch 0 taken 137154357 times.

✓ Branch 1 taken 3344873 times.

✗ Branch 2 not taken.

✓ Branch 3 taken 103615568 times.

✓ Branch 4 taken 3342732 times.

247457530

switch (cs->levels_for_compare) {

5103

137154357

case 1:

5104

137154357

return my_strnxfrm_uca_900_tmpl<Mb_wc_utf8mb4, 1>(

5105

137153061

cs, Mb_wc_utf8mb4(), dst, dstlen, src, srclen, flags);

5106

3344873

case 2:

5107

3344873

return my_strnxfrm_uca_900_tmpl<Mb_wc_utf8mb4, 2>(

5108

3344873

cs, Mb_wc_utf8mb4(), dst, dstlen, src, srclen, flags);

5109

✗

default:

5110

✗

assert(false);

5111

103615568

case 3:

5112

103615568

return my_strnxfrm_uca_900_tmpl<Mb_wc_utf8mb4, 3>(

5113

103615568

cs, Mb_wc_utf8mb4(), dst, dstlen, src, srclen, flags);

5114

3342732

case 4:

5115

3342732

return my_strnxfrm_uca_900_tmpl<Mb_wc_utf8mb4, 4>(

5116

3342738

cs, Mb_wc_utf8mb4(), dst, dstlen, src, srclen, flags);

5117

}

5118

} else {

5119

✗

Mb_wc_through_function_pointer mb_wc(cs);

5120

✗

switch (cs->levels_for_compare) {

5121

✗

case 1:

5122

✗

return my_strnxfrm_uca_900_tmpl<decltype(mb_wc), 1>(

5123

✗

cs, mb_wc, dst, dstlen, src, srclen, flags);

5124

✗

case 2:

5125

✗

return my_strnxfrm_uca_900_tmpl<decltype(mb_wc), 2>(

5126

✗

cs, mb_wc, dst, dstlen, src, srclen, flags);

5127

✗

default:

5128

✗

assert(false);

5129

✗

case 3:

5130

✗

return my_strnxfrm_uca_900_tmpl<decltype(mb_wc), 3>(

5131

✗

cs, mb_wc, dst, dstlen, src, srclen, flags);

5132

✗

case 4:

5133

✗

return my_strnxfrm_uca_900_tmpl<decltype(mb_wc), 4>(

5134

✗

cs, mb_wc, dst, dstlen, src, srclen, flags);

}

}

}

1303790

static size_t my_strnxfrmlen_uca_900(const CHARSET_INFO *cs, size_t len) {

5140

/*

5141

The character with the most weights is U+FDFA ARABIC LIGATURE SALLALLAHOU

5142

ALAYHE WASALLAM, which we truncate to eight weights. This is the most we

5143

can get in regular DUCET.

5144

5145

In addition, collations with reorderings can add an extra weight per weight,

5146

which currently only happens on the primary level. We simulate this by

5147

simply adding an extra level.

5148

5149

One could conceivably have tailorings yielding expansions having more than

5150

this, but we don't currently, and mostly, tailorings are about contractions

5151

and adding single weights anyway.

5152

5153

We also need to add room for one level separator between each level.

5154

*/

5155

// We really ought to have len % 4 == 0, but not all calling code conforms.

5156

1303790

const size_t num_codepoints = (len + 3) / 4;

5157

1303790

const size_t max_num_weights_per_level = num_codepoints * 8;

5158

1303790

size_t max_num_weights = max_num_weights_per_level * cs->levels_for_compare;

5159

4/4

✓ Branch 0 taken 798 times.

✓ Branch 1 taken 1302992 times.

✓ Branch 2 taken 638 times.

✓ Branch 3 taken 160 times.

1303790

if (cs->coll_param && cs->coll_param->reorder_param) {

5160

638

max_num_weights += max_num_weights_per_level;

5161

}

5162

1303790

return (max_num_weights + (cs->levels_for_compare - 1)) * sizeof(uint16_t);

}

} // extern "C"

/*

UCS2 optimized CHARSET_INFO compatible wrappers.

5169

*/

5170

extern "C" {

5171

12

static int my_strnncoll_ucs2_uca(const CHARSET_INFO *cs, const uchar *s,

5172

size_t slen, const uchar *t, size_t tlen,

5173

bool t_is_prefix) {

5174

12

Mb_wc_through_function_pointer mb_wc(cs);

5175

1/2

✓ Branch 0 taken 12 times.

✗ Branch 1 not taken.

12

return my_strnncoll_uca<uca_scanner_any<decltype(mb_wc)>, 1>(

5176

24

cs, mb_wc, s, slen, t, tlen, t_is_prefix);

5177

}

5178

5179

12139

static int my_strnncollsp_ucs2_uca(const CHARSET_INFO *cs, const uchar *s,

5180

size_t slen, const uchar *t, size_t tlen) {

5181

12139

Mb_wc_through_function_pointer mb_wc(cs);

5182

1/2

✓ Branch 0 taken 12139 times.

✗ Branch 1 not taken.

24278

return my_strnncollsp_uca(cs, mb_wc, s, slen, t, tlen);

5183

}

5184

5185

460

static void my_hash_sort_ucs2_uca(const CHARSET_INFO *cs, const uchar *s,

5186

size_t slen, uint64 *n1, uint64 *n2) {

5187

460

Mb_wc_through_function_pointer mb_wc(cs);

5188

1/2

✓ Branch 0 taken 460 times.

✗ Branch 1 not taken.

460

my_hash_sort_uca(cs, mb_wc, s, slen, n1, n2);

5189

460

}

5190

5191

4869686

static size_t my_strnxfrm_ucs2_uca(const CHARSET_INFO *cs, uchar *dst,

5192

size_t dstlen, uint num_codepoints,

5193

const uchar *src, size_t srclen,

5194

uint flags) {

5195

4869686

Mb_wc_through_function_pointer mb_wc(cs);

5196

1/2

✓ Branch 0 taken 4869686 times.

✗ Branch 1 not taken.

4869686

return my_strnxfrm_uca(cs, mb_wc, dst, dstlen, num_codepoints, src, srclen,

5197

9739372

flags);

}

} // extern "C"

MY_COLLATION_HANDLER my_collation_ucs2_uca_handler = {

5202

my_coll_init_uca, /* init */

5203

my_coll_uninit_uca,

5204

my_strnncoll_ucs2_uca,

5205

my_strnncollsp_ucs2_uca,

5206

my_strnxfrm_ucs2_uca,

5207

my_strnxfrmlen_simple,

5208

my_like_range_generic,

my_wildcmp_uca,

nullptr,

my_instr_mb,

my_hash_sort_ucs2_uca,

5213

my_propagate_complex};

5214

5215

CHARSET_INFO my_charset_ucs2_unicode_ci = {

128,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5220

"ucs2", /* cs name */

5221

"ucs2_unicode_ci", /* m_coll_name */

5222

"UCS-2 Unicode", /* comment */

5223

"", /* tailoring */

5224

nullptr, /* coll_param */

5225

nullptr, /* ctype */

5226

nullptr, /* to_lower */

5227

nullptr, /* to_upper */

5228

nullptr, /* sort_order */

5229

nullptr, /* uca */

5230

nullptr, /* tab_to_uni */

5231

nullptr, /* tab_from_uni */

5232

&my_unicase_default, /* caseinfo */

5233

nullptr, /* state_map */

5234

nullptr, /* ident_map */

5235

8, /* strxfrm_multiply */

5236

1, /* caseup_multiply */

5237

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5242

0xFFFF, /* max_sort_char */

5243

' ', /* pad char */

5244

false, /* escape_with_backslash_is_dangerous */

5245

1, /* levels_for_compare */

5246

&my_charset_ucs2_handler,

5247

&my_collation_ucs2_uca_handler,

5248

PAD_SPACE};

5249

5250

CHARSET_INFO my_charset_ucs2_icelandic_uca_ci = {

129,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5255

"ucs2", /* cs name */

5256

"ucs2_icelandic_ci", /* m_coll_name */

5257

"UCS-2 Unicode", /* comment */

5258

icelandic, /* tailoring */

5259

nullptr, /* coll_param */

5260

nullptr, /* ctype */

5261

nullptr, /* to_lower */

5262

nullptr, /* to_upper */

5263

nullptr, /* sort_order */

5264

nullptr, /* uca */

5265

nullptr, /* tab_to_uni */

5266

nullptr, /* tab_from_uni */

5267

&my_unicase_default, /* caseinfo */

5268

nullptr, /* state_map */

5269

nullptr, /* ident_map */

5270

8, /* strxfrm_multiply */

5271

1, /* caseup_multiply */

5272

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5277

0xFFFF, /* max_sort_char */

5278

' ', /* pad char */

5279

false, /* escape_with_backslash_is_dangerous */

5280

1, /* levels_for_compare */

5281

&my_charset_ucs2_handler,

5282

&my_collation_ucs2_uca_handler,

5283

PAD_SPACE};

5284

5285

CHARSET_INFO my_charset_ucs2_latvian_uca_ci = {

130,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5290

"ucs2", /* cs name */

5291

"ucs2_latvian_ci", /* m_coll_name */

5292

"UCS-2 Unicode", /* comment */

5293

latvian, /* tailoring */

5294

nullptr, /* coll_param */

5295

nullptr, /* ctype */

5296

nullptr, /* to_lower */

5297

nullptr, /* to_upper */

5298

nullptr, /* sort_order */

5299

nullptr, /* uca */

5300

nullptr, /* tab_to_uni */

5301

nullptr, /* tab_from_uni */

5302

&my_unicase_default, /* caseinfo */

5303

nullptr, /* state_map */

5304

nullptr, /* ident_map */

5305

8, /* strxfrm_multiply */

5306

1, /* caseup_multiply */

5307

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5312

0xFFFF, /* max_sort_char */

5313

' ', /* pad char */

5314

false, /* escape_with_backslash_is_dangerous */

5315

1, /* levels_for_compare */

5316

&my_charset_ucs2_handler,

5317

&my_collation_ucs2_uca_handler,

5318

PAD_SPACE};

5319

5320

CHARSET_INFO my_charset_ucs2_romanian_uca_ci = {

131,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5325

"ucs2", /* cs name */

5326

"ucs2_romanian_ci", /* m_coll_name */

5327

"UCS-2 Unicode", /* comment */

5328

romanian, /* tailoring */

5329

nullptr, /* coll_param */

5330

nullptr, /* ctype */

5331

nullptr, /* to_lower */

5332

nullptr, /* to_upper */

5333

nullptr, /* sort_order */

5334

nullptr, /* uca */

5335

nullptr, /* tab_to_uni */

5336

nullptr, /* tab_from_uni */

5337

&my_unicase_default, /* caseinfo */

5338

nullptr, /* state_map */

5339

nullptr, /* ident_map */

5340

8, /* strxfrm_multiply */

5341

1, /* caseup_multiply */

5342

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5347

0xFFFF, /* max_sort_char */

5348

' ', /* pad char */

5349

false, /* escape_with_backslash_is_dangerous */

5350

1, /* levels_for_compare */

5351

&my_charset_ucs2_handler,

5352

&my_collation_ucs2_uca_handler,

5353

PAD_SPACE};

5354

5355

CHARSET_INFO my_charset_ucs2_slovenian_uca_ci = {

132,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5360

"ucs2", /* cs name */

5361

"ucs2_slovenian_ci", /* m_coll_name */

5362

"UCS-2 Unicode", /* comment */

5363

slovenian, /* tailoring */

5364

nullptr, /* coll_param */

5365

nullptr, /* ctype */

5366

nullptr, /* to_lower */

5367

nullptr, /* to_upper */

5368

nullptr, /* sort_order */

5369

nullptr, /* uca */

5370

nullptr, /* tab_to_uni */

5371

nullptr, /* tab_from_uni */

5372

&my_unicase_default, /* caseinfo */

5373

nullptr, /* state_map */

5374

nullptr, /* ident_map */

5375

8, /* strxfrm_multiply */

5376

1, /* caseup_multiply */

5377

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5382

0xFFFF, /* max_sort_char */

5383

' ', /* pad char */

5384

false, /* escape_with_backslash_is_dangerous */

5385

1, /* levels_for_compare */

5386

&my_charset_ucs2_handler,

5387

&my_collation_ucs2_uca_handler,

5388

PAD_SPACE};

5389

5390

CHARSET_INFO my_charset_ucs2_polish_uca_ci = {

133,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5395

"ucs2", /* cs name */

5396

"ucs2_polish_ci", /* m_coll_name */

5397

"UCS-2 Unicode", /* comment */

5398

polish, /* tailoring */

5399

nullptr, /* coll_param */

5400

nullptr, /* ctype */

5401

nullptr, /* to_lower */

5402

nullptr, /* to_upper */

5403

nullptr, /* sort_order */

5404

nullptr, /* uca */

5405

nullptr, /* tab_to_uni */

5406

nullptr, /* tab_from_uni */

5407

&my_unicase_default, /* caseinfo */

5408

nullptr, /* state_map */

5409

nullptr, /* ident_map */

5410

8, /* strxfrm_multiply */

5411

1, /* caseup_multiply */

5412

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5417

0xFFFF, /* max_sort_char */

5418

' ', /* pad char */

5419

false, /* escape_with_backslash_is_dangerous */

5420

1, /* levels_for_compare */

5421

&my_charset_ucs2_handler,

5422

&my_collation_ucs2_uca_handler,

5423

PAD_SPACE};

5424

5425

CHARSET_INFO my_charset_ucs2_estonian_uca_ci = {

134,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5430

"ucs2", /* cs name */

5431

"ucs2_estonian_ci", /* m_coll_name */

5432

"UCS-2 Unicode", /* comment */

5433

estonian, /* tailoring */

5434

nullptr, /* coll_param */

5435

nullptr, /* ctype */

5436

nullptr, /* to_lower */

5437

nullptr, /* to_upper */

5438

nullptr, /* sort_order */

5439

nullptr, /* uca */

5440

nullptr, /* tab_to_uni */

5441

nullptr, /* tab_from_uni */

5442

&my_unicase_default, /* caseinfo */

5443

nullptr, /* state_map */

5444

nullptr, /* ident_map */

5445

8, /* strxfrm_multiply */

5446

1, /* caseup_multiply */

5447

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5452

0xFFFF, /* max_sort_char */

5453

' ', /* pad char */

5454

false, /* escape_with_backslash_is_dangerous */

5455

1, /* levels_for_compare */

5456

&my_charset_ucs2_handler,

5457

&my_collation_ucs2_uca_handler,

5458

PAD_SPACE};

5459

5460

CHARSET_INFO my_charset_ucs2_spanish_uca_ci = {

135,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5465

"ucs2", /* cs name */

5466

"ucs2_spanish_ci", /* m_coll_name */

5467

"UCS-2 Unicode", /* comment */

5468

spanish, /* tailoring */

5469

nullptr, /* coll_param */

5470

nullptr, /* ctype */

5471

nullptr, /* to_lower */

5472

nullptr, /* to_upper */

5473

nullptr, /* sort_order */

5474

nullptr, /* uca */

5475

nullptr, /* tab_to_uni */

5476

nullptr, /* tab_from_uni */

5477

&my_unicase_default, /* caseinfo */

5478

nullptr, /* state_map */

5479

nullptr, /* ident_map */

5480

8, /* strxfrm_multiply */

5481

1, /* caseup_multiply */

5482

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5487

0xFFFF, /* max_sort_char */

5488

' ', /* pad char */

5489

false, /* escape_with_backslash_is_dangerous */

5490

1, /* levels_for_compare */

5491

&my_charset_ucs2_handler,

5492

&my_collation_ucs2_uca_handler,

5493

PAD_SPACE};

5494

5495

CHARSET_INFO my_charset_ucs2_swedish_uca_ci = {

136,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5500

"ucs2", /* cs name */

5501

"ucs2_swedish_ci", /* m_coll_name */

5502

"UCS-2 Unicode", /* comment */

5503

swedish, /* tailoring */

5504

nullptr, /* coll_param */

5505

nullptr, /* ctype */

5506

nullptr, /* to_lower */

5507

nullptr, /* to_upper */

5508

nullptr, /* sort_order */

5509

nullptr, /* uca */

5510

nullptr, /* tab_to_uni */

5511

nullptr, /* tab_from_uni */

5512

&my_unicase_default, /* caseinfo */

5513

nullptr, /* state_map */

5514

nullptr, /* ident_map */

5515

8, /* strxfrm_multiply */

5516

1, /* caseup_multiply */

5517

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5522

0xFFFF, /* max_sort_char */

5523

' ', /* pad char */

5524

false, /* escape_with_backslash_is_dangerous */

5525

1, /* levels_for_compare */

5526

&my_charset_ucs2_handler,

5527

&my_collation_ucs2_uca_handler,

5528

PAD_SPACE};

5529

5530

CHARSET_INFO my_charset_ucs2_turkish_uca_ci = {

137,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5535

"ucs2", /* cs name */

5536

"ucs2_turkish_ci", /* m_coll_name */

5537

"UCS-2 Unicode", /* comment */

5538

turkish, /* tailoring */

5539

nullptr, /* coll_param */

5540

nullptr, /* ctype */

5541

nullptr, /* to_lower */

5542

nullptr, /* to_upper */

5543

nullptr, /* sort_order */

5544

nullptr, /* uca */

5545

nullptr, /* tab_to_uni */

5546

nullptr, /* tab_from_uni */

5547

&my_unicase_turkish, /* caseinfo */

5548

nullptr, /* state_map */

5549

nullptr, /* ident_map */

5550

8, /* strxfrm_multiply */

5551

1, /* caseup_multiply */

5552

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5557

0xFFFF, /* max_sort_char */

5558

' ', /* pad char */

5559

false, /* escape_with_backslash_is_dangerous */

5560

1, /* levels_for_compare */

5561

&my_charset_ucs2_handler,

5562

&my_collation_ucs2_uca_handler,

5563

PAD_SPACE};

5564

5565

CHARSET_INFO my_charset_ucs2_czech_uca_ci = {

138,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5570

"ucs2", /* cs name */

5571

"ucs2_czech_ci", /* m_coll_name */

5572

"UCS-2 Unicode", /* comment */

5573

czech, /* tailoring */

5574

nullptr, /* coll_param */

5575

nullptr, /* ctype */

5576

nullptr, /* to_lower */

5577

nullptr, /* to_upper */

5578

nullptr, /* sort_order */

5579

nullptr, /* uca */

5580

nullptr, /* tab_to_uni */

5581

nullptr, /* tab_from_uni */

5582

&my_unicase_default, /* caseinfo */

5583

nullptr, /* state_map */

5584

nullptr, /* ident_map */

5585

8, /* strxfrm_multiply */

5586

1, /* caseup_multiply */

5587

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5592

0xFFFF, /* max_sort_char */

5593

' ', /* pad char */

5594

false, /* escape_with_backslash_is_dangerous */

5595

1, /* levels_for_compare */

5596

&my_charset_ucs2_handler,

5597

&my_collation_ucs2_uca_handler,

5598

PAD_SPACE};

5599

5600

CHARSET_INFO my_charset_ucs2_danish_uca_ci = {

139,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5605

"ucs2", /* cs name */

5606

"ucs2_danish_ci", /* m_coll_name */

5607

"UCS-2 Unicode", /* comment */

5608

danish, /* tailoring */

5609

nullptr, /* coll_param */

5610

nullptr, /* ctype */

5611

nullptr, /* to_lower */

5612

nullptr, /* to_upper */

5613

nullptr, /* sort_order */

5614

nullptr, /* uca */

5615

nullptr, /* tab_to_uni */

5616

nullptr, /* tab_from_uni */

5617

&my_unicase_default, /* caseinfo */

5618

nullptr, /* state_map */

5619

nullptr, /* ident_map */

5620

8, /* strxfrm_multiply */

5621

1, /* caseup_multiply */

5622

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5627

0xFFFF, /* max_sort_char */

5628

' ', /* pad char */

5629

false, /* escape_with_backslash_is_dangerous */

5630

1, /* levels_for_compare */

5631

&my_charset_ucs2_handler,

5632

&my_collation_ucs2_uca_handler,

5633

PAD_SPACE};

5634

5635

CHARSET_INFO my_charset_ucs2_lithuanian_uca_ci = {

140,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5640

"ucs2", /* cs name */

5641

"ucs2_lithuanian_ci", /* m_coll_name */

5642

"UCS-2 Unicode", /* comment */

5643

lithuanian, /* tailoring */

5644

nullptr, /* coll_param */

5645

nullptr, /* ctype */

5646

nullptr, /* to_lower */

5647

nullptr, /* to_upper */

5648

nullptr, /* sort_order */

5649

nullptr, /* uca */

5650

nullptr, /* tab_to_uni */

5651

nullptr, /* tab_from_uni */

5652

&my_unicase_default, /* caseinfo */

5653

nullptr, /* state_map */

5654

nullptr, /* ident_map */

5655

8, /* strxfrm_multiply */

5656

1, /* caseup_multiply */

5657

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5662

0xFFFF, /* max_sort_char */

5663

' ', /* pad char */

5664

false, /* escape_with_backslash_is_dangerous */

5665

1, /* levels_for_compare */

5666

&my_charset_ucs2_handler,

5667

&my_collation_ucs2_uca_handler,

5668

PAD_SPACE};

5669

5670

CHARSET_INFO my_charset_ucs2_slovak_uca_ci = {

141,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5675

"ucs2", /* cs name */

5676

"ucs2_slovak_ci", /* m_coll_name */

5677

"UCS-2 Unicode", /* comment */

5678

slovak, /* tailoring */

5679

nullptr, /* coll_param */

5680

nullptr, /* ctype */

5681

nullptr, /* to_lower */

5682

nullptr, /* to_upper */

5683

nullptr, /* sort_order */

5684

nullptr, /* uca */

5685

nullptr, /* tab_to_uni */

5686

nullptr, /* tab_from_uni */

5687

&my_unicase_default, /* caseinfo */

5688

nullptr, /* state_map */

5689

nullptr, /* ident_map */

5690

8, /* strxfrm_multiply */

5691

1, /* caseup_multiply */

5692

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5697

0xFFFF, /* max_sort_char */

5698

' ', /* pad char */

5699

false, /* escape_with_backslash_is_dangerous */

5700

1, /* levels_for_compare */

5701

&my_charset_ucs2_handler,

5702

&my_collation_ucs2_uca_handler,

5703

PAD_SPACE};

5704

5705

CHARSET_INFO my_charset_ucs2_spanish2_uca_ci = {

142,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5710

"ucs2", /* cs name */

5711

"ucs2_spanish2_ci", /* m_coll_name */

5712

"UCS-2 Unicode", /* comment */

5713

spanish2, /* tailoring */

5714

nullptr, /* coll_param */

5715

nullptr, /* ctype */

5716

nullptr, /* to_lower */

5717

nullptr, /* to_upper */

5718

nullptr, /* sort_order */

5719

nullptr, /* uca */

5720

nullptr, /* tab_to_uni */

5721

nullptr, /* tab_from_uni */

5722

&my_unicase_default, /* caseinfo */

5723

nullptr, /* state_map */

5724

nullptr, /* ident_map */

5725

8, /* strxfrm_multiply */

5726

1, /* caseup_multiply */

5727

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5732

0xFFFF, /* max_sort_char */

5733

' ', /* pad char */

5734

false, /* escape_with_backslash_is_dangerous */

5735

1, /* levels_for_compare */

5736

&my_charset_ucs2_handler,

5737

&my_collation_ucs2_uca_handler,

5738

PAD_SPACE};

5739

5740

CHARSET_INFO my_charset_ucs2_roman_uca_ci = {

143,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5745

"ucs2", /* cs name */

5746

"ucs2_roman_ci", /* m_coll_name */

5747

"UCS-2 Unicode", /* comment */

5748

roman, /* tailoring */

5749

nullptr, /* coll_param */

5750

nullptr, /* ctype */

5751

nullptr, /* to_lower */

5752

nullptr, /* to_upper */

5753

nullptr, /* sort_order */

5754

nullptr, /* uca */

5755

nullptr, /* tab_to_uni */

5756

nullptr, /* tab_from_uni */

5757

&my_unicase_default, /* caseinfo */

5758

nullptr, /* state_map */

5759

nullptr, /* ident_map */

5760

8, /* strxfrm_multiply */

5761

1, /* caseup_multiply */

5762

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5767

0xFFFF, /* max_sort_char */

5768

' ', /* pad char */

5769

false, /* escape_with_backslash_is_dangerous */

5770

1, /* levels_for_compare */

5771

&my_charset_ucs2_handler,

5772

&my_collation_ucs2_uca_handler,

5773

PAD_SPACE};

5774

5775

CHARSET_INFO my_charset_ucs2_persian_uca_ci = {

144,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5780

"ucs2", /* cs name */

5781

"ucs2_persian_ci", /* m_coll_name */

5782

"UCS-2 Unicode", /* comment */

5783

persian, /* tailoring */

5784

nullptr, /* coll_param */

5785

nullptr, /* ctype */

5786

nullptr, /* to_lower */

5787

nullptr, /* to_upper */

5788

nullptr, /* sort_order */

5789

nullptr, /* uca */

5790

nullptr, /* tab_to_uni */

5791

nullptr, /* tab_from_uni */

5792

&my_unicase_default, /* caseinfo */

5793

nullptr, /* state_map */

5794

nullptr, /* ident_map */

5795

8, /* strxfrm_multiply */

5796

1, /* caseup_multiply */

5797

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5802

0xFFFF, /* max_sort_char */

5803

' ', /* pad char */

5804

false, /* escape_with_backslash_is_dangerous */

5805

1, /* levels_for_compare */

5806

&my_charset_ucs2_handler,

5807

&my_collation_ucs2_uca_handler,

5808

PAD_SPACE};

5809

5810

CHARSET_INFO my_charset_ucs2_esperanto_uca_ci = {

145,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5815

"ucs2", /* cs name */

5816

"ucs2_esperanto_ci", /* m_coll_name */

5817

"UCS-2 Unicode", /* comment */

5818

esperanto, /* tailoring */

5819

nullptr, /* coll_param */

5820

nullptr, /* ctype */

5821

nullptr, /* to_lower */

5822

nullptr, /* to_upper */

5823

nullptr, /* sort_order */

5824

nullptr, /* uca */

5825

nullptr, /* tab_to_uni */

5826

nullptr, /* tab_from_uni */

5827

&my_unicase_default, /* caseinfo */

5828

nullptr, /* state_map */

5829

nullptr, /* ident_map */

5830

8, /* strxfrm_multiply */

5831

1, /* caseup_multiply */

5832

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5837

0xFFFF, /* max_sort_char */

5838

' ', /* pad char */

5839

false, /* escape_with_backslash_is_dangerous */

5840

1, /* levels_for_compare */

5841

&my_charset_ucs2_handler,

5842

&my_collation_ucs2_uca_handler,

5843

PAD_SPACE};

5844

5845

CHARSET_INFO my_charset_ucs2_hungarian_uca_ci = {

146,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5850

"ucs2", /* cs name */

5851

"ucs2_hungarian_ci", /* m_coll_name */

5852

"UCS-2 Unicode", /* comment */

5853

hungarian, /* tailoring */

5854

nullptr, /* coll_param */

5855

nullptr, /* ctype */

5856

nullptr, /* to_lower */

5857

nullptr, /* to_upper */

5858

nullptr, /* sort_order */

5859

nullptr, /* uca */

5860

nullptr, /* tab_to_uni */

5861

nullptr, /* tab_from_uni */

5862

&my_unicase_default, /* caseinfo */

5863

nullptr, /* state_map */

5864

nullptr, /* ident_map */

5865

8, /* strxfrm_multiply */

5866

1, /* caseup_multiply */

5867

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5872

0xFFFF, /* max_sort_char */

5873

' ', /* pad char */

5874

false, /* escape_with_backslash_is_dangerous */

5875

1, /* levels_for_compare */

5876

&my_charset_ucs2_handler,

5877

&my_collation_ucs2_uca_handler,

5878

PAD_SPACE};

5879

5880

CHARSET_INFO my_charset_ucs2_sinhala_uca_ci = {

147,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5885

"ucs2", /* csname */

5886

"ucs2_sinhala_ci", /* m_coll_name */

5887

"UCS-2 Unicode", /* comment */

5888

sinhala, /* tailoring */

5889

nullptr, /* coll_param */

5890

nullptr, /* ctype */

5891

nullptr, /* to_lower */

5892

nullptr, /* to_upper */

5893

nullptr, /* sort_order */

5894

nullptr, /* uca */

5895

nullptr, /* tab_to_uni */

5896

nullptr, /* tab_from_uni */

5897

&my_unicase_default, /* caseinfo */

5898

nullptr, /* state_map */

5899

nullptr, /* ident_map */

5900

8, /* strxfrm_multiply */

5901

1, /* caseup_multiply */

5902

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5907

0xFFFF, /* max_sort_char */

5908

' ', /* pad char */

5909

false, /* escape_with_backslash_is_dangerous */

5910

1, /* levels_for_compare */

5911

&my_charset_ucs2_handler,

5912

&my_collation_ucs2_uca_handler,

5913

PAD_SPACE};

5914

5915

CHARSET_INFO my_charset_ucs2_german2_uca_ci = {

148,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5920

"ucs2", /* csname */

5921

"ucs2_german2_ci", /* m_coll_name */

5922

"UCS-2 Unicode", /* comment */

5923

german2, /* tailoring */

5924

nullptr, /* coll_param */

5925

nullptr, /* ctype */

5926

nullptr, /* to_lower */

5927

nullptr, /* to_upper */

5928

nullptr, /* sort_order */

5929

nullptr, /* uca */

5930

nullptr, /* tab_to_uni */

5931

nullptr, /* tab_from_uni */

5932

&my_unicase_default, /* caseinfo */

5933

nullptr, /* state_map */

5934

nullptr, /* ident_map */

5935

8, /* strxfrm_multiply */

5936

1, /* caseup_multiply */

5937

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5942

0xFFFF, /* max_sort_char */

5943

' ', /* pad char */

5944

false, /* escape_with_backslash_is_dangerous */

5945

1, /* levels_for_compare */

5946

&my_charset_ucs2_handler,

5947

&my_collation_ucs2_uca_handler,

5948

PAD_SPACE};

5949

5950

CHARSET_INFO my_charset_ucs2_croatian_uca_ci = {

149,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5955

"ucs2", /* csname */

5956

"ucs2_croatian_ci", /* m_coll_name */

5957

"UCS-2 Unicode", /* comment */

5958

croatian, /* tailoring */

5959

nullptr, /* coll_param */

5960

nullptr, /* ctype */

5961

nullptr, /* to_lower */

5962

nullptr, /* to_upper */

5963

nullptr, /* sort_order */

5964

nullptr, /* uca */

5965

nullptr, /* tab_to_uni */

5966

nullptr, /* tab_from_uni */

5967

&my_unicase_default, /* caseinfo */

5968

nullptr, /* state_map */

5969

nullptr, /* ident_map */

5970

8, /* strxfrm_multiply */

5971

1, /* caseup_multiply */

5972

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

5977

0xFFFF, /* max_sort_char */

5978

' ', /* pad char */

5979

false, /* escape_with_backslash_is_dangerous */

5980

1, /* levels_for_compare */

5981

&my_charset_ucs2_handler,

5982

&my_collation_ucs2_uca_handler,

5983

PAD_SPACE};

5984

5985

CHARSET_INFO my_charset_ucs2_unicode_520_ci = {

150,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

5990

"ucs2", /* cs name */

5991

"ucs2_unicode_520_ci", /* m_coll_name */

5992

"UCS-2 Unicode", /* comment */

5993

"", /* tailoring */

5994

nullptr, /* coll_param */

5995

nullptr, /* ctype */

5996

nullptr, /* to_lower */

5997

nullptr, /* to_upper */

5998

nullptr, /* sort_order */

5999

&my_uca_v520, /* uca */

6000

nullptr, /* tab_to_uni */

6001

nullptr, /* tab_from_uni */

6002

&my_unicase_unicode520, /* caseinfo */

6003

nullptr, /* state_map */

6004

nullptr, /* ident_map */

6005

8, /* strxfrm_multiply */

6006

1, /* caseup_multiply */

6007

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6012

0xFFFF, /* max_sort_char */

6013

' ', /* pad char */

6014

false, /* escape_with_backslash_is_dangerous */

6015

1, /* levels_for_compare */

6016

&my_charset_ucs2_handler,

6017

&my_collation_ucs2_uca_handler,

6018

PAD_SPACE};

6019

6020

CHARSET_INFO my_charset_ucs2_vietnamese_ci = {

151,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

6025

"ucs2", /* csname */

6026

"ucs2_vietnamese_ci", /* m_coll_name */

6027

"UCS-2 Unicode", /* comment */

6028

vietnamese, /* tailoring */

6029

nullptr, /* coll_param */

6030

nullptr, /* ctype */

6031

nullptr, /* to_lower */

6032

nullptr, /* to_upper */

6033

nullptr, /* sort_order */

6034

nullptr, /* uca */

6035

nullptr, /* tab_to_uni */

6036

nullptr, /* tab_from_uni */

6037

&my_unicase_default, /* caseinfo */

6038

nullptr, /* state_map */

6039

nullptr, /* ident_map */

6040

8, /* strxfrm_multiply */

6041

1, /* caseup_multiply */

6042

1, /* casedn_multiply */

2, /* mbminlen */

2, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6047

0xFFFF, /* max_sort_char */

6048

' ', /* pad char */

6049

false, /* escape_with_backslash_is_dangerous */

6050

1, /* levels_for_compare */

6051

&my_charset_ucs2_handler,

6052

&my_collation_ucs2_uca_handler,

6053

PAD_SPACE};

6054

6055

MY_COLLATION_HANDLER my_collation_any_uca_handler = {

6056

my_coll_init_uca, /* init */

6057

my_coll_uninit_uca, my_strnncoll_any_uca, my_strnncollsp_any_uca,

6058

my_strnxfrm_any_uca, my_strnxfrmlen_simple, my_like_range_mb,

6059

my_wildcmp_uca, my_strcasecmp_uca, my_instr_mb,

6060

my_hash_sort_any_uca, my_propagate_complex};

6061

6062

MY_COLLATION_HANDLER my_collation_uca_900_handler = {

6063

my_coll_init_uca, /* init */

6064

my_coll_uninit_uca, my_strnncoll_uca_900, my_strnncollsp_uca_900,

6065

my_strnxfrm_uca_900, my_strnxfrmlen_uca_900, my_like_range_mb,

6066

my_wildcmp_uca, my_strcasecmp_uca, my_instr_mb,

6067

my_hash_sort_uca_900, my_propagate_uca_900};

6068

6069

/*

6070

We consider bytes with code more than 127 as a letter.

6071

This guarantees that word boundaries work fine with regular

6072

expressions. Note, there is no need to mark byte 255 as a

6073

letter, it is illegal byte in UTF8.

6074

*/

6075

static const uchar ctype_utf8[] = {

6076

0, 32, 32, 32, 32, 32, 32, 32, 32, 32, 40, 40, 40, 40, 40, 32,

6077

32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32,

6078

32, 72, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16,

6079

16, 132, 132, 132, 132, 132, 132, 132, 132, 132, 132, 16, 16, 16, 16, 16,

6080

16, 16, 129, 129, 129, 129, 129, 129, 1, 1, 1, 1, 1, 1, 1, 1,

6081

1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 16, 16, 16, 16,

6082

16, 16, 130, 130, 130, 130, 130, 130, 2, 2, 2, 2, 2, 2, 2, 2,

6083

2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 16, 16, 16, 16,

6084

32, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6085

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6086

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6087

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6088

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6089

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6090

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6091

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

6092

0};

6093

6094

extern MY_CHARSET_HANDLER my_charset_utf8_handler;

6095

6096

#define MY_CS_UTF8MB3_UCA_FLAGS \

6097

(MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE)

6098

6099

CHARSET_INFO my_charset_utf8_unicode_ci = {

192,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6104

"utf8mb3", /* cs name */

6105

"utf8mb3_unicode_ci", /* m_coll_name */

6106

"UCS-2 Unicode", /* comment */

6107

"", /* tailoring */

6108

nullptr, /* coll_param */

6109

ctype_utf8, /* ctype */

6110

nullptr, /* to_lower */

6111

nullptr, /* to_upper */

6112

nullptr, /* sort_order */

6113

nullptr, /* uca */

6114

nullptr, /* tab_to_uni */

6115

nullptr, /* tab_from_uni */

6116

&my_unicase_default, /* caseinfo */

6117

nullptr, /* state_map */

6118

nullptr, /* ident_map */

6119

8, /* strxfrm_multiply */

6120

1, /* caseup_multiply */

6121

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6126

0xFFFF, /* max_sort_char */

6127

' ', /* pad char */

6128

false, /* escape_with_backslash_is_dangerous */

6129

1, /* levels_for_compare */

6130

&my_charset_utf8_handler,

6131

&my_collation_any_uca_handler,

6132

PAD_SPACE};

6133

6134

CHARSET_INFO my_charset_utf8_icelandic_uca_ci = {

193,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6139

"utf8mb3", /* cs name */

6140

"utf8mb3_icelandic_ci", /* m_coll_name */

6141

"UTF-8 Unicode", /* comment */

6142

icelandic, /* tailoring */

6143

nullptr, /* coll_param */

6144

ctype_utf8, /* ctype */

6145

nullptr, /* to_lower */

6146

nullptr, /* to_upper */

6147

nullptr, /* sort_order */

6148

nullptr, /* uca */

6149

nullptr, /* tab_to_uni */

6150

nullptr, /* tab_from_uni */

6151

&my_unicase_default, /* caseinfo */

6152

nullptr, /* state_map */

6153

nullptr, /* ident_map */

6154

8, /* strxfrm_multiply */

6155

1, /* caseup_multiply */

6156

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6161

0xFFFF, /* max_sort_char */

6162

' ', /* pad char */

6163

false, /* escape_with_backslash_is_dangerous */

6164

1, /* levels_for_compare */

6165

&my_charset_utf8_handler,

6166

&my_collation_any_uca_handler,

6167

PAD_SPACE};

6168

6169

CHARSET_INFO my_charset_utf8_latvian_uca_ci = {

194,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6174

"utf8mb3", /* cs name */

6175

"utf8mb3_latvian_ci", /* m_coll_name */

6176

"UTF-8 Unicode", /* comment */

6177

latvian, /* tailoring */

6178

nullptr, /* coll_param */

6179

ctype_utf8, /* ctype */

6180

nullptr, /* to_lower */

6181

nullptr, /* to_upper */

6182

nullptr, /* sort_order */

6183

nullptr, /* uca */

6184

nullptr, /* tab_to_uni */

6185

nullptr, /* tab_from_uni */

6186

&my_unicase_default, /* caseinfo */

6187

nullptr, /* state_map */

6188

nullptr, /* ident_map */

6189

8, /* strxfrm_multiply */

6190

1, /* caseup_multiply */

6191

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6196

0xFFFF, /* max_sort_char */

6197

' ', /* pad char */

6198

false, /* escape_with_backslash_is_dangerous */

6199

1, /* levels_for_compare */

6200

&my_charset_utf8_handler,

6201

&my_collation_any_uca_handler,

6202

PAD_SPACE};

6203

6204

CHARSET_INFO my_charset_utf8_romanian_uca_ci = {

195,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6209

"utf8mb3", /* cs name */

6210

"utf8mb3_romanian_ci", /* m_coll_name */

6211

"UTF-8 Unicode", /* comment */

6212

romanian, /* tailoring */

6213

nullptr, /* coll_param */

6214

ctype_utf8, /* ctype */

6215

nullptr, /* to_lower */

6216

nullptr, /* to_upper */

6217

nullptr, /* sort_order */

6218

nullptr, /* uca */

6219

nullptr, /* tab_to_uni */

6220

nullptr, /* tab_from_uni */

6221

&my_unicase_default, /* caseinfo */

6222

nullptr, /* state_map */

6223

nullptr, /* ident_map */

6224

8, /* strxfrm_multiply */

6225

1, /* caseup_multiply */

6226

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6231

0xFFFF, /* max_sort_char */

6232

' ', /* pad char */

6233

false, /* escape_with_backslash_is_dangerous */

6234

1, /* levels_for_compare */

6235

&my_charset_utf8_handler,

6236

&my_collation_any_uca_handler,

6237

PAD_SPACE};

6238

6239

CHARSET_INFO my_charset_utf8_slovenian_uca_ci = {

196,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6244

"utf8mb3", /* cs name */

6245

"utf8mb3_slovenian_ci", /* m_coll_name */

6246

"UTF-8 Unicode", /* comment */

6247

slovenian, /* tailoring */

6248

nullptr, /* coll_param */

6249

ctype_utf8, /* ctype */

6250

nullptr, /* to_lower */

6251

nullptr, /* to_upper */

6252

nullptr, /* sort_order */

6253

nullptr, /* uca */

6254

nullptr, /* tab_to_uni */

6255

nullptr, /* tab_from_uni */

6256

&my_unicase_default, /* caseinfo */

6257

nullptr, /* state_map */

6258

nullptr, /* ident_map */

6259

8, /* strxfrm_multiply */

6260

1, /* caseup_multiply */

6261

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6266

0xFFFF, /* max_sort_char */

6267

' ', /* pad char */

6268

false, /* escape_with_backslash_is_dangerous */

6269

1, /* levels_for_compare */

6270

&my_charset_utf8_handler,

6271

&my_collation_any_uca_handler,

6272

PAD_SPACE};

6273

6274

CHARSET_INFO my_charset_utf8_polish_uca_ci = {

197,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6279

"utf8mb3", /* cs name */

6280

"utf8mb3_polish_ci", /* m_coll_name */

6281

"UTF-8 Unicode", /* comment */

6282

polish, /* tailoring */

6283

nullptr, /* coll_param */

6284

ctype_utf8, /* ctype */

6285

nullptr, /* to_lower */

6286

nullptr, /* to_upper */

6287

nullptr, /* sort_order */

6288

nullptr, /* uca */

6289

nullptr, /* tab_to_uni */

6290

nullptr, /* tab_from_uni */

6291

&my_unicase_default, /* caseinfo */

6292

nullptr, /* state_map */

6293

nullptr, /* ident_map */

6294

8, /* strxfrm_multiply */

6295

1, /* caseup_multiply */

6296

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6301

0xFFFF, /* max_sort_char */

6302

' ', /* pad char */

6303

false, /* escape_with_backslash_is_dangerous */

6304

1, /* levels_for_compare */

6305

&my_charset_utf8_handler,

6306

&my_collation_any_uca_handler,

6307

PAD_SPACE};

6308

6309

CHARSET_INFO my_charset_utf8_estonian_uca_ci = {

198,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6314

"utf8mb3", /* cs name */

6315

"utf8mb3_estonian_ci", /* m_coll_name */

6316

"UTF-8 Unicode", /* comment */

6317

estonian, /* tailoring */

6318

nullptr, /* coll_param */

6319

ctype_utf8, /* ctype */

6320

nullptr, /* to_lower */

6321

nullptr, /* to_upper */

6322

nullptr, /* sort_order */

6323

nullptr, /* uca */

6324

nullptr, /* tab_to_uni */

6325

nullptr, /* tab_from_uni */

6326

&my_unicase_default, /* caseinfo */

6327

nullptr, /* state_map */

6328

nullptr, /* ident_map */

6329

8, /* strxfrm_multiply */

6330

1, /* caseup_multiply */

6331

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6336

0xFFFF, /* max_sort_char */

6337

' ', /* pad char */

6338

false, /* escape_with_backslash_is_dangerous */

6339

1, /* levels_for_compare */

6340

&my_charset_utf8_handler,

6341

&my_collation_any_uca_handler,

6342

PAD_SPACE};

6343

6344

CHARSET_INFO my_charset_utf8_spanish_uca_ci = {

199,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6349

"utf8mb3", /* cs name */

6350

"utf8mb3_spanish_ci", /* m_coll_name */

6351

"UTF-8 Unicode", /* comment */

6352

spanish, /* tailoring */

6353

nullptr, /* coll_param */

6354

ctype_utf8, /* ctype */

6355

nullptr, /* to_lower */

6356

nullptr, /* to_upper */

6357

nullptr, /* sort_order */

6358

nullptr, /* uca */

6359

nullptr, /* tab_to_uni */

6360

nullptr, /* tab_from_uni */

6361

&my_unicase_default, /* caseinfo */

6362

nullptr, /* state_map */

6363

nullptr, /* ident_map */

6364

8, /* strxfrm_multiply */

6365

1, /* caseup_multiply */

6366

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6371

0xFFFF, /* max_sort_char */

6372

' ', /* pad char */

6373

false, /* escape_with_backslash_is_dangerous */

6374

1, /* levels_for_compare */

6375

&my_charset_utf8_handler,

6376

&my_collation_any_uca_handler,

6377

PAD_SPACE};

6378

6379

CHARSET_INFO my_charset_utf8_swedish_uca_ci = {

200,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6384

"utf8mb3", /* cs name */

6385

"utf8mb3_swedish_ci", /* m_coll_name */

6386

"UTF-8 Unicode", /* comment */

6387

swedish, /* tailoring */

6388

nullptr, /* coll_param */

6389

ctype_utf8, /* ctype */

6390

nullptr, /* to_lower */

6391

nullptr, /* to_upper */

6392

nullptr, /* sort_order */

6393

nullptr, /* uca */

6394

nullptr, /* tab_to_uni */

6395

nullptr, /* tab_from_uni */

6396

&my_unicase_default, /* caseinfo */

6397

nullptr, /* state_map */

6398

nullptr, /* ident_map */

6399

8, /* strxfrm_multiply */

6400

1, /* caseup_multiply */

6401

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6406

0xFFFF, /* max_sort_char */

6407

' ', /* pad char */

6408

false, /* escape_with_backslash_is_dangerous */

6409

1, /* levels_for_compare */

6410

&my_charset_utf8_handler,

6411

&my_collation_any_uca_handler,

6412

PAD_SPACE};

6413

6414

CHARSET_INFO my_charset_utf8_turkish_uca_ci = {

201,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6419

"utf8mb3", /* cs name */

6420

"utf8mb3_turkish_ci", /* m_coll_name */

6421

"UTF-8 Unicode", /* comment */

6422

turkish, /* tailoring */

6423

nullptr, /* coll_param */

6424

ctype_utf8, /* ctype */

6425

nullptr, /* to_lower */

6426

nullptr, /* to_upper */

6427

nullptr, /* sort_order */

6428

nullptr, /* uca */

6429

nullptr, /* tab_to_uni */

6430

nullptr, /* tab_from_uni */

6431

&my_unicase_turkish, /* caseinfo */

6432

nullptr, /* state_map */

6433

nullptr, /* ident_map */

6434

8, /* strxfrm_multiply */

6435

2, /* caseup_multiply */

6436

2, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6441

0xFFFF, /* max_sort_char */

6442

' ', /* pad char */

6443

false, /* escape_with_backslash_is_dangerous */

6444

1, /* levels_for_compare */

6445

&my_charset_utf8_handler,

6446

&my_collation_any_uca_handler,

6447

PAD_SPACE};

6448

6449

CHARSET_INFO my_charset_utf8_czech_uca_ci = {

202,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6454

"utf8mb3", /* cs name */

6455

"utf8mb3_czech_ci", /* m_coll_name */

6456

"UTF-8 Unicode", /* comment */

6457

czech, /* tailoring */

6458

nullptr, /* coll_param */

6459

ctype_utf8, /* ctype */

6460

nullptr, /* to_lower */

6461

nullptr, /* to_upper */

6462

nullptr, /* sort_order */

6463

nullptr, /* uca */

6464

nullptr, /* tab_to_uni */

6465

nullptr, /* tab_from_uni */

6466

&my_unicase_default, /* caseinfo */

6467

nullptr, /* state_map */

6468

nullptr, /* ident_map */

6469

8, /* strxfrm_multiply */

6470

1, /* caseup_multiply */

6471

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6476

0xFFFF, /* max_sort_char */

6477

' ', /* pad char */

6478

false, /* escape_with_backslash_is_dangerous */

6479

1, /* levels_for_compare */

6480

&my_charset_utf8_handler,

6481

&my_collation_any_uca_handler,

6482

PAD_SPACE};

6483

6484

CHARSET_INFO my_charset_utf8_danish_uca_ci = {

203,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6489

"utf8mb3", /* cs name */

6490

"utf8mb3_danish_ci", /* m_coll_name */

6491

"UTF-8 Unicode", /* comment */

6492

danish, /* tailoring */

6493

nullptr, /* coll_param */

6494

ctype_utf8, /* ctype */

6495

nullptr, /* to_lower */

6496

nullptr, /* to_upper */

6497

nullptr, /* sort_order */

6498

nullptr, /* uca */

6499

nullptr, /* tab_to_uni */

6500

nullptr, /* tab_from_uni */

6501

&my_unicase_default, /* caseinfo */

6502

nullptr, /* state_map */

6503

nullptr, /* ident_map */

6504

8, /* strxfrm_multiply */

6505

1, /* caseup_multiply */

6506

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6511

0xFFFF, /* max_sort_char */

6512

' ', /* pad char */

6513

false, /* escape_with_backslash_is_dangerous */

6514

1, /* levels_for_compare */

6515

&my_charset_utf8_handler,

6516

&my_collation_any_uca_handler,

6517

PAD_SPACE};

6518

6519

CHARSET_INFO my_charset_utf8_lithuanian_uca_ci = {

204,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6524

"utf8mb3", /* cs name */

6525

"utf8mb3_lithuanian_ci", /* m_coll_name */

6526

"UTF-8 Unicode", /* comment */

6527

lithuanian, /* tailoring */

6528

nullptr, /* coll_param */

6529

ctype_utf8, /* ctype */

6530

nullptr, /* to_lower */

6531

nullptr, /* to_upper */

6532

nullptr, /* sort_order */

6533

nullptr, /* uca */

6534

nullptr, /* tab_to_uni */

6535

nullptr, /* tab_from_uni */

6536

&my_unicase_default, /* caseinfo */

6537

nullptr, /* state_map */

6538

nullptr, /* ident_map */

6539

8, /* strxfrm_multiply */

6540

1, /* caseup_multiply */

6541

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6546

0xFFFF, /* max_sort_char */

6547

' ', /* pad char */

6548

false, /* escape_with_backslash_is_dangerous */

6549

1, /* levels_for_compare */

6550

&my_charset_utf8_handler,

6551

&my_collation_any_uca_handler,

6552

PAD_SPACE};

6553

6554

CHARSET_INFO my_charset_utf8_slovak_uca_ci = {

205,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6559

"utf8mb3", /* cs name */

6560

"utf8mb3_slovak_ci", /* m_coll_name */

6561

"UTF-8 Unicode", /* comment */

6562

slovak, /* tailoring */

6563

nullptr, /* coll_param */

6564

ctype_utf8, /* ctype */

6565

nullptr, /* to_lower */

6566

nullptr, /* to_upper */

6567

nullptr, /* sort_order */

6568

nullptr, /* uca */

6569

nullptr, /* tab_to_uni */

6570

nullptr, /* tab_from_uni */

6571

&my_unicase_default, /* caseinfo */

6572

nullptr, /* state_map */

6573

nullptr, /* ident_map */

6574

8, /* strxfrm_multiply */

6575

1, /* caseup_multiply */

6576

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6581

0xFFFF, /* max_sort_char */

6582

' ', /* pad char */

6583

false, /* escape_with_backslash_is_dangerous */

6584

1, /* levels_for_compare */

6585

&my_charset_utf8_handler,

6586

&my_collation_any_uca_handler,

6587

PAD_SPACE};

6588

6589

CHARSET_INFO my_charset_utf8_spanish2_uca_ci = {

206,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6594

"utf8mb3", /* cs name */

6595

"utf8mb3_spanish2_ci", /* m_coll_name */

6596

"UTF-8 Unicode", /* comment */

6597

spanish2, /* tailoring */

6598

nullptr, /* coll_param */

6599

ctype_utf8, /* ctype */

6600

nullptr, /* to_lower */

6601

nullptr, /* to_upper */

6602

nullptr, /* sort_order */

6603

nullptr, /* uca */

6604

nullptr, /* tab_to_uni */

6605

nullptr, /* tab_from_uni */

6606

&my_unicase_default, /* caseinfo */

6607

nullptr, /* state_map */

6608

nullptr, /* ident_map */

6609

8, /* strxfrm_multiply */

6610

1, /* caseup_multiply */

6611

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6616

0xFFFF, /* max_sort_char */

6617

' ', /* pad char */

6618

false, /* escape_with_backslash_is_dangerous */

6619

1, /* levels_for_compare */

6620

&my_charset_utf8_handler,

6621

&my_collation_any_uca_handler,

6622

PAD_SPACE};

6623

6624

CHARSET_INFO my_charset_utf8_roman_uca_ci = {

207,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6629

"utf8mb3", /* cs name */

6630

"utf8mb3_roman_ci", /* m_coll_name */

6631

"UTF-8 Unicode", /* comment */

6632

roman, /* tailoring */

6633

nullptr, /* coll_param */

6634

ctype_utf8, /* ctype */

6635

nullptr, /* to_lower */

6636

nullptr, /* to_upper */

6637

nullptr, /* sort_order */

6638

nullptr, /* uca */

6639

nullptr, /* tab_to_uni */

6640

nullptr, /* tab_from_uni */

6641

&my_unicase_default, /* caseinfo */

6642

nullptr, /* state_map */

6643

nullptr, /* ident_map */

6644

8, /* strxfrm_multiply */

6645

1, /* caseup_multiply */

6646

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6651

0xFFFF, /* max_sort_char */

6652

' ', /* pad char */

6653

false, /* escape_with_backslash_is_dangerous */

6654

1, /* levels_for_compare */

6655

&my_charset_utf8_handler,

6656

&my_collation_any_uca_handler,

6657

PAD_SPACE};

6658

6659

CHARSET_INFO my_charset_utf8_persian_uca_ci = {

208,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6664

"utf8mb3", /* cs name */

6665

"utf8mb3_persian_ci", /* m_coll_name */

6666

"UTF-8 Unicode", /* comment */

6667

persian, /* tailoring */

6668

nullptr, /* coll_param */

6669

ctype_utf8, /* ctype */

6670

nullptr, /* to_lower */

6671

nullptr, /* to_upper */

6672

nullptr, /* sort_order */

6673

nullptr, /* uca */

6674

nullptr, /* tab_to_uni */

6675

nullptr, /* tab_from_uni */

6676

&my_unicase_default, /* caseinfo */

6677

nullptr, /* state_map */

6678

nullptr, /* ident_map */

6679

8, /* strxfrm_multiply */

6680

1, /* caseup_multiply */

6681

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6686

0xFFFF, /* max_sort_char */

6687

' ', /* pad char */

6688

false, /* escape_with_backslash_is_dangerous */

6689

1, /* levels_for_compare */

6690

&my_charset_utf8_handler,

6691

&my_collation_any_uca_handler,

6692

PAD_SPACE};

6693

6694

CHARSET_INFO my_charset_utf8_esperanto_uca_ci = {

209,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6699

"utf8mb3", /* cs name */

6700

"utf8mb3_esperanto_ci", /* m_coll_name */

6701

"UTF-8 Unicode", /* comment */

6702

esperanto, /* tailoring */

6703

nullptr, /* coll_param */

6704

ctype_utf8, /* ctype */

6705

nullptr, /* to_lower */

6706

nullptr, /* to_upper */

6707

nullptr, /* sort_order */

6708

nullptr, /* uca */

6709

nullptr, /* tab_to_uni */

6710

nullptr, /* tab_from_uni */

6711

&my_unicase_default, /* caseinfo */

6712

nullptr, /* state_map */

6713

nullptr, /* ident_map */

6714

8, /* strxfrm_multiply */

6715

1, /* caseup_multiply */

6716

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6721

0xFFFF, /* max_sort_char */

6722

' ', /* pad char */

6723

false, /* escape_with_backslash_is_dangerous */

6724

1, /* levels_for_compare */

6725

&my_charset_utf8_handler,

6726

&my_collation_any_uca_handler,

6727

PAD_SPACE};

6728

6729

CHARSET_INFO my_charset_utf8_hungarian_uca_ci = {

210,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6734

"utf8mb3", /* cs name */

6735

"utf8mb3_hungarian_ci", /* m_coll_name */

6736

"UTF-8 Unicode", /* comment */

6737

hungarian, /* tailoring */

6738

nullptr, /* coll_param */

6739

ctype_utf8, /* ctype */

6740

nullptr, /* to_lower */

6741

nullptr, /* to_upper */

6742

nullptr, /* sort_order */

6743

nullptr, /* uca */

6744

nullptr, /* tab_to_uni */

6745

nullptr, /* tab_from_uni */

6746

&my_unicase_default, /* caseinfo */

6747

nullptr, /* state_map */

6748

nullptr, /* ident_map */

6749

8, /* strxfrm_multiply */

6750

1, /* caseup_multiply */

6751

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6756

0xFFFF, /* max_sort_char */

6757

' ', /* pad char */

6758

false, /* escape_with_backslash_is_dangerous */

6759

1, /* levels_for_compare */

6760

&my_charset_utf8_handler,

6761

&my_collation_any_uca_handler,

6762

PAD_SPACE};

6763

6764

CHARSET_INFO my_charset_utf8_sinhala_uca_ci = {

211,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6769

"utf8mb3", /* cs name */

6770

"utf8mb3_sinhala_ci", /* m_coll_name */

6771

"UTF-8 Unicode", /* comment */

6772

sinhala, /* tailoring */

6773

nullptr, /* coll_param */

6774

ctype_utf8, /* ctype */

6775

nullptr, /* to_lower */

6776

nullptr, /* to_upper */

6777

nullptr, /* sort_order */

6778

nullptr, /* uca */

6779

nullptr, /* tab_to_uni */

6780

nullptr, /* tab_from_uni */

6781

&my_unicase_default, /* caseinfo */

6782

nullptr, /* state_map */

6783

nullptr, /* ident_map */

6784

8, /* strxfrm_multiply */

6785

1, /* caseup_multiply */

6786

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6791

0xFFFF, /* max_sort_char */

6792

' ', /* pad char */

6793

false, /* escape_with_backslash_is_dangerous */

6794

1, /* levels_for_compare */

6795

&my_charset_utf8_handler,

6796

&my_collation_any_uca_handler,

6797

PAD_SPACE};

6798

6799

CHARSET_INFO my_charset_utf8_german2_uca_ci = {

212,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6804

"utf8mb3", /* cs name */

6805

"utf8mb3_german2_ci", /* m_coll_name */

6806

"UTF-8 Unicode", /* comment */

6807

german2, /* tailoring */

6808

nullptr, /* coll_param */

6809

ctype_utf8, /* ctype */

6810

nullptr, /* to_lower */

6811

nullptr, /* to_upper */

6812

nullptr, /* sort_order */

6813

nullptr, /* uca */

6814

nullptr, /* tab_to_uni */

6815

nullptr, /* tab_from_uni */

6816

&my_unicase_default, /* caseinfo */

6817

nullptr, /* state_map */

6818

nullptr, /* ident_map */

6819

8, /* strxfrm_multiply */

6820

1, /* caseup_multiply */

6821

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6826

0xFFFF, /* max_sort_char */

6827

' ', /* pad char */

6828

false, /* escape_with_backslash_is_dangerous */

6829

1, /* levels_for_compare */

6830

&my_charset_utf8_handler,

6831

&my_collation_any_uca_handler,

6832

PAD_SPACE};

6833

6834

CHARSET_INFO my_charset_utf8_croatian_uca_ci = {

213,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6839

"utf8mb3", /* cs name */

6840

"utf8mb3_croatian_ci", /* m_coll_name */

6841

"UTF-8 Unicode", /* comment */

6842

croatian, /* tailoring */

6843

nullptr, /* coll_param */

6844

ctype_utf8, /* ctype */

6845

nullptr, /* to_lower */

6846

nullptr, /* to_upper */

6847

nullptr, /* sort_order */

6848

nullptr, /* uca */

6849

nullptr, /* tab_to_uni */

6850

nullptr, /* tab_from_uni */

6851

&my_unicase_default, /* caseinfo */

6852

nullptr, /* state_map */

6853

nullptr, /* ident_map */

6854

8, /* strxfrm_multiply */

6855

1, /* caseup_multiply */

6856

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6861

0xFFFF, /* max_sort_char */

6862

' ', /* pad char */

6863

false, /* escape_with_backslash_is_dangerous */

6864

1, /* levels_for_compare */

6865

&my_charset_utf8_handler,

6866

&my_collation_any_uca_handler,

6867

PAD_SPACE};

6868

6869

CHARSET_INFO my_charset_utf8_unicode_520_ci = {

214,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6874

"utf8mb3", /* csname */

6875

"utf8mb3_unicode_520_ci", /* m_coll_name */

6876

"UTF-8 Unicode", /* comment */

6877

"", /* tailoring */

6878

nullptr, /* coll_param */

6879

ctype_utf8, /* ctype */

6880

nullptr, /* to_lower */

6881

nullptr, /* to_upper */

6882

nullptr, /* sort_order */

6883

&my_uca_v520, /* uca */

6884

nullptr, /* tab_to_uni */

6885

nullptr, /* tab_from_uni */

6886

&my_unicase_unicode520, /* caseinfo */

6887

nullptr, /* state_map */

6888

nullptr, /* ident_map */

6889

8, /* strxfrm_multiply */

6890

1, /* caseup_multiply */

6891

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6896

0xFFFF, /* max_sort_char */

6897

' ', /* pad char */

6898

false, /* escape_with_backslash_is_dangerous */

6899

1, /* levels_for_compare */

6900

&my_charset_utf8_handler,

6901

&my_collation_any_uca_handler,

6902

PAD_SPACE};

6903

6904

CHARSET_INFO my_charset_utf8_vietnamese_ci = {

215,

0,

0, /* number */

MY_CS_UTF8MB3_UCA_FLAGS, /* flags */

6909

"utf8mb3", /* cs name */

6910

"utf8mb3_vietnamese_ci", /* m_coll_name */

6911

"UTF-8 Unicode", /* comment */

6912

vietnamese, /* tailoring */

6913

nullptr, /* coll_param */

6914

ctype_utf8, /* ctype */

6915

nullptr, /* to_lower */

6916

nullptr, /* to_upper */

6917

nullptr, /* sort_order */

6918

nullptr, /* uca */

6919

nullptr, /* tab_to_uni */

6920

nullptr, /* tab_from_uni */

6921

&my_unicase_default, /* caseinfo */

6922

nullptr, /* state_map */

6923

nullptr, /* ident_map */

6924

8, /* strxfrm_multiply */

6925

1, /* caseup_multiply */

6926

1, /* casedn_multiply */

1, /* mbminlen */

3, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6931

0xFFFF, /* max_sort_char */

6932

' ', /* pad char */

6933

false, /* escape_with_backslash_is_dangerous */

6934

1, /* levels_for_compare */

6935

&my_charset_utf8_handler,

6936

&my_collation_any_uca_handler,

6937

PAD_SPACE};

6938

6939

extern MY_CHARSET_HANDLER my_charset_utf8mb4_handler;

6940

6941

#define MY_CS_UTF8MB4_UCA_FLAGS \

6942

(MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_UNICODE_SUPPLEMENT)

6943

6944

CHARSET_INFO my_charset_utf8mb4_unicode_ci = {

224,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

6949

MY_UTF8MB4, /* csname */

6950

MY_UTF8MB4 "_unicode_ci", /* m_coll_name */

6951

"UTF-8 Unicode", /* comment */

6952

"", /* tailoring */

6953

nullptr, /* coll_param */

6954

ctype_utf8, /* ctype */

6955

nullptr, /* to_lower */

6956

nullptr, /* to_upper */

6957

nullptr, /* sort_order */

6958

nullptr, /* uca */

6959

nullptr, /* tab_to_uni */

6960

nullptr, /* tab_from_uni */

6961

&my_unicase_default, /* caseinfo */

6962

nullptr, /* state_map */

6963

nullptr, /* ident_map */

6964

8, /* strxfrm_multiply */

6965

1, /* caseup_multiply */

6966

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

6971

0xFFFF, /* max_sort_char */

6972

' ', /* pad char */

6973

false, /* escape_with_backslash_is_dangerous */

6974

1, /* levels_for_compare */

6975

&my_charset_utf8mb4_handler,

6976

&my_collation_any_uca_handler,

6977

PAD_SPACE};

6978

6979

CHARSET_INFO my_charset_utf8mb4_icelandic_uca_ci = {

225,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

6984

MY_UTF8MB4, /* csname */

6985

MY_UTF8MB4 "_icelandic_ci", /* m_coll_name */

6986

"UTF-8 Unicode", /* comment */

6987

icelandic, /* tailoring */

6988

nullptr, /* coll_param */

6989

ctype_utf8, /* ctype */

6990

nullptr, /* to_lower */

6991

nullptr, /* to_upper */

6992

nullptr, /* sort_order */

6993

nullptr, /* uca */

6994

nullptr, /* tab_to_uni */

6995

nullptr, /* tab_from_uni */

6996

&my_unicase_default, /* caseinfo */

6997

nullptr, /* state_map */

6998

nullptr, /* ident_map */

6999

8, /* strxfrm_multiply */

7000

1, /* caseup_multiply */

7001

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7006

0xFFFF, /* max_sort_char */

7007

' ', /* pad char */

7008

false, /* escape_with_backslash_is_dangerous */

7009

1, /* levels_for_compare */

7010

&my_charset_utf8mb4_handler,

7011

&my_collation_any_uca_handler,

7012

PAD_SPACE};

7013

7014

CHARSET_INFO my_charset_utf8mb4_latvian_uca_ci = {

226,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7019

MY_UTF8MB4, /* csname */

7020

MY_UTF8MB4 "_latvian_ci", /* m_coll_name */

7021

"UTF-8 Unicode", /* comment */

7022

latvian, /* tailoring */

7023

nullptr, /* coll_param */

7024

ctype_utf8, /* ctype */

7025

nullptr, /* to_lower */

7026

nullptr, /* to_upper */

7027

nullptr, /* sort_order */

7028

nullptr, /* uca */

7029

nullptr, /* tab_to_uni */

7030

nullptr, /* tab_from_uni */

7031

&my_unicase_default, /* caseinfo */

7032

nullptr, /* state_map */

7033

nullptr, /* ident_map */

7034

8, /* strxfrm_multiply */

7035

1, /* caseup_multiply */

7036

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7041

0xFFFF, /* max_sort_char */

7042

' ', /* pad char */

7043

false, /* escape_with_backslash_is_dangerous */

7044

1, /* levels_for_compare */

7045

&my_charset_utf8mb4_handler,

7046

&my_collation_any_uca_handler,

7047

PAD_SPACE};

7048

7049

CHARSET_INFO my_charset_utf8mb4_romanian_uca_ci = {

227,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7054

MY_UTF8MB4, /* csname */

7055

MY_UTF8MB4 "_romanian_ci", /* m_coll_name */

7056

"UTF-8 Unicode", /* comment */

7057

romanian, /* tailoring */

7058

nullptr, /* coll_param */

7059

ctype_utf8, /* ctype */

7060

nullptr, /* to_lower */

7061

nullptr, /* to_upper */

7062

nullptr, /* sort_order */

7063

nullptr, /* uca */

7064

nullptr, /* tab_to_uni */

7065

nullptr, /* tab_from_uni */

7066

&my_unicase_default, /* caseinfo */

7067

nullptr, /* state_map */

7068

nullptr, /* ident_map */

7069

8, /* strxfrm_multiply */

7070

1, /* caseup_multiply */

7071

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7076

0xFFFF, /* max_sort_char */

7077

' ', /* pad char */

7078

false, /* escape_with_backslash_is_dangerous */

7079

1, /* levels_for_compare */

7080

&my_charset_utf8mb4_handler,

7081

&my_collation_any_uca_handler,

7082

PAD_SPACE};

7083

7084

CHARSET_INFO my_charset_utf8mb4_slovenian_uca_ci = {

228,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7089

MY_UTF8MB4, /* csname */

7090

MY_UTF8MB4 "_slovenian_ci", /* m_coll_name */

7091

"UTF-8 Unicode", /* comment */

7092

slovenian, /* tailoring */

7093

nullptr, /* coll_param */

7094

ctype_utf8, /* ctype */

7095

nullptr, /* to_lower */

7096

nullptr, /* to_upper */

7097

nullptr, /* sort_order */

7098

nullptr, /* uca */

7099

nullptr, /* tab_to_uni */

7100

nullptr, /* tab_from_uni */

7101

&my_unicase_default, /* caseinfo */

7102

nullptr, /* state_map */

7103

nullptr, /* ident_map */

7104

8, /* strxfrm_multiply */

7105

1, /* caseup_multiply */

7106

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7111

0xFFFF, /* max_sort_char */

7112

' ', /* pad char */

7113

false, /* escape_with_backslash_is_dangerous */

7114

1, /* levels_for_compare */

7115

&my_charset_utf8mb4_handler,

7116

&my_collation_any_uca_handler,

7117

PAD_SPACE};

7118

7119

CHARSET_INFO my_charset_utf8mb4_polish_uca_ci = {

229,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7124

MY_UTF8MB4, /* csname */

7125

MY_UTF8MB4 "_polish_ci", /* m_coll_name */

7126

"UTF-8 Unicode", /* comment */

7127

polish, /* tailoring */

7128

nullptr, /* coll_param */

7129

ctype_utf8, /* ctype */

7130

nullptr, /* to_lower */

7131

nullptr, /* to_upper */

7132

nullptr, /* sort_order */

7133

nullptr, /* uca */

7134

nullptr, /* tab_to_uni */

7135

nullptr, /* tab_from_uni */

7136

&my_unicase_default, /* caseinfo */

7137

nullptr, /* state_map */

7138

nullptr, /* ident_map */

7139

8, /* strxfrm_multiply */

7140

1, /* caseup_multiply */

7141

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7146

0xFFFF, /* max_sort_char */

7147

' ', /* pad char */

7148

false, /* escape_with_backslash_is_dangerous */

7149

1, /* levels_for_compare */

7150

&my_charset_utf8mb4_handler,

7151

&my_collation_any_uca_handler,

7152

PAD_SPACE};

7153

7154

CHARSET_INFO my_charset_utf8mb4_estonian_uca_ci = {

230,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7159

MY_UTF8MB4, /* csname */

7160

MY_UTF8MB4 "_estonian_ci", /* m_coll_name */

7161

"UTF-8 Unicode", /* comment */

7162

estonian, /* tailoring */

7163

nullptr, /* coll_param */

7164

ctype_utf8, /* ctype */

7165

nullptr, /* to_lower */

7166

nullptr, /* to_upper */

7167

nullptr, /* sort_order */

7168

nullptr, /* uca */

7169

nullptr, /* tab_to_uni */

7170

nullptr, /* tab_from_uni */

7171

&my_unicase_default, /* caseinfo */

7172

nullptr, /* state_map */

7173

nullptr, /* ident_map */

7174

8, /* strxfrm_multiply */

7175

1, /* caseup_multiply */

7176

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7181

0xFFFF, /* max_sort_char */

7182

' ', /* pad char */

7183

false, /* escape_with_backslash_is_dangerous */

7184

1, /* levels_for_compare */

7185

&my_charset_utf8mb4_handler,

7186

&my_collation_any_uca_handler,

7187

PAD_SPACE};

7188

7189

CHARSET_INFO my_charset_utf8mb4_spanish_uca_ci = {

231,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7194

MY_UTF8MB4, /* csname */

7195

MY_UTF8MB4 "_spanish_ci", /* m_coll_name */

7196

"UTF-8 Unicode", /* comment */

7197

spanish, /* tailoring */

7198

nullptr, /* coll_param */

7199

ctype_utf8, /* ctype */

7200

nullptr, /* to_lower */

7201

nullptr, /* to_upper */

7202

nullptr, /* sort_order */

7203

nullptr, /* uca */

7204

nullptr, /* tab_to_uni */

7205

nullptr, /* tab_from_uni */

7206

&my_unicase_default, /* caseinfo */

7207

nullptr, /* state_map */

7208

nullptr, /* ident_map */

7209

8, /* strxfrm_multiply */

7210

1, /* caseup_multiply */

7211

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7216

0xFFFF, /* max_sort_char */

7217

' ', /* pad char */

7218

false, /* escape_with_backslash_is_dangerous */

7219

1, /* levels_for_compare */

7220

&my_charset_utf8mb4_handler,

7221

&my_collation_any_uca_handler,

7222

PAD_SPACE};

7223

7224

CHARSET_INFO my_charset_utf8mb4_swedish_uca_ci = {

232,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7229

MY_UTF8MB4, /* csname */

7230

MY_UTF8MB4 "_swedish_ci", /* m_coll_name */

7231

"UTF-8 Unicode", /* comment */

7232

swedish, /* tailoring */

7233

nullptr, /* coll_param */

7234

ctype_utf8, /* ctype */

7235

nullptr, /* to_lower */

7236

nullptr, /* to_upper */

7237

nullptr, /* sort_order */

7238

nullptr, /* uca */

7239

nullptr, /* tab_to_uni */

7240

nullptr, /* tab_from_uni */

7241

&my_unicase_default, /* caseinfo */

7242

nullptr, /* state_map */

7243

nullptr, /* ident_map */

7244

8, /* strxfrm_multiply */

7245

1, /* caseup_multiply */

7246

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7251

0xFFFF, /* max_sort_char */

7252

' ', /* pad char */

7253

false, /* escape_with_backslash_is_dangerous */

7254

1, /* levels_for_compare */

7255

&my_charset_utf8mb4_handler,

7256

&my_collation_any_uca_handler,

7257

PAD_SPACE};

7258

7259

CHARSET_INFO my_charset_utf8mb4_turkish_uca_ci = {

233,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7264

MY_UTF8MB4, /* csname */

7265

MY_UTF8MB4 "_turkish_ci", /* m_coll_name */

7266

"UTF-8 Unicode", /* comment */

7267

turkish, /* tailoring */

7268

nullptr, /* coll_param */

7269

ctype_utf8, /* ctype */

7270

nullptr, /* to_lower */

7271

nullptr, /* to_upper */

7272

nullptr, /* sort_order */

7273

nullptr, /* uca */

7274

nullptr, /* tab_to_uni */

7275

nullptr, /* tab_from_uni */

7276

&my_unicase_turkish, /* caseinfo */

7277

nullptr, /* state_map */

7278

nullptr, /* ident_map */

7279

8, /* strxfrm_multiply */

7280

2, /* caseup_multiply */

7281

2, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7286

0xFFFF, /* max_sort_char */

7287

' ', /* pad char */

7288

false, /* escape_with_backslash_is_dangerous */

7289

1, /* levels_for_compare */

7290

&my_charset_utf8mb4_handler,

7291

&my_collation_any_uca_handler,

7292

PAD_SPACE};

7293

7294

CHARSET_INFO my_charset_utf8mb4_czech_uca_ci = {

234,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7299

MY_UTF8MB4, /* csname */

7300

MY_UTF8MB4 "_czech_ci", /* m_coll_name */

7301

"UTF-8 Unicode", /* comment */

7302

czech, /* tailoring */

7303

nullptr, /* coll_param */

7304

ctype_utf8, /* ctype */

7305

nullptr, /* to_lower */

7306

nullptr, /* to_upper */

7307

nullptr, /* sort_order */

7308

nullptr, /* uca */

7309

nullptr, /* tab_to_uni */

7310

nullptr, /* tab_from_uni */

7311

&my_unicase_default, /* caseinfo */

7312

nullptr, /* state_map */

7313

nullptr, /* ident_map */

7314

8, /* strxfrm_multiply */

7315

1, /* caseup_multiply */

7316

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7321

0xFFFF, /* max_sort_char */

7322

' ', /* pad char */

7323

false, /* escape_with_backslash_is_dangerous */

7324

1, /* levels_for_compare */

7325

&my_charset_utf8mb4_handler,

7326

&my_collation_any_uca_handler,

7327

PAD_SPACE};

7328

7329

CHARSET_INFO my_charset_utf8mb4_danish_uca_ci = {

235,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7334

MY_UTF8MB4, /* csname */

7335

MY_UTF8MB4 "_danish_ci", /* m_coll_name */

7336

"UTF-8 Unicode", /* comment */

7337

danish, /* tailoring */

7338

nullptr, /* coll_param */

7339

ctype_utf8, /* ctype */

7340

nullptr, /* to_lower */

7341

nullptr, /* to_upper */

7342

nullptr, /* sort_order */

7343

nullptr, /* uca */

7344

nullptr, /* tab_to_uni */

7345

nullptr, /* tab_from_uni */

7346

&my_unicase_default, /* caseinfo */

7347

nullptr, /* state_map */

7348

nullptr, /* ident_map */

7349

8, /* strxfrm_multiply */

7350

1, /* caseup_multiply */

7351

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7356

0xFFFF, /* max_sort_char */

7357

' ', /* pad char */

7358

false, /* escape_with_backslash_is_dangerous */

7359

1, /* levels_for_compare */

7360

&my_charset_utf8mb4_handler,

7361

&my_collation_any_uca_handler,

7362

PAD_SPACE};

7363

7364

CHARSET_INFO my_charset_utf8mb4_lithuanian_uca_ci = {

236,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7369

MY_UTF8MB4, /* csname */

7370

MY_UTF8MB4 "_lithuanian_ci", /* m_coll_name */

7371

"UTF-8 Unicode", /* comment */

7372

lithuanian, /* tailoring */

7373

nullptr, /* coll_param */

7374

ctype_utf8, /* ctype */

7375

nullptr, /* to_lower */

7376

nullptr, /* to_upper */

7377

nullptr, /* sort_order */

7378

nullptr, /* uca */

7379

nullptr, /* tab_to_uni */

7380

nullptr, /* tab_from_uni */

7381

&my_unicase_default, /* caseinfo */

7382

nullptr, /* state_map */

7383

nullptr, /* ident_map */

7384

8, /* strxfrm_multiply */

7385

1, /* caseup_multiply */

7386

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7391

0xFFFF, /* max_sort_char */

7392

' ', /* pad char */

7393

false, /* escape_with_backslash_is_dangerous */

7394

1, /* levels_for_compare */

7395

&my_charset_utf8mb4_handler,

7396

&my_collation_any_uca_handler,

7397

PAD_SPACE};

7398

7399

CHARSET_INFO my_charset_utf8mb4_slovak_uca_ci = {

237,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7404

MY_UTF8MB4, /* csname */

7405

MY_UTF8MB4 "_slovak_ci", /* m_coll_name */

7406

"UTF-8 Unicode", /* comment */

7407

slovak, /* tailoring */

7408

nullptr, /* coll_param */

7409

ctype_utf8, /* ctype */

7410

nullptr, /* to_lower */

7411

nullptr, /* to_upper */

7412

nullptr, /* sort_order */

7413

nullptr, /* uca */

7414

nullptr, /* tab_to_uni */

7415

nullptr, /* tab_from_uni */

7416

&my_unicase_default, /* caseinfo */

7417

nullptr, /* state_map */

7418

nullptr, /* ident_map */

7419

8, /* strxfrm_multiply */

7420

1, /* caseup_multiply */

7421

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7426

0xFFFF, /* max_sort_char */

7427

' ', /* pad char */

7428

false, /* escape_with_backslash_is_dangerous */

7429

1, /* levels_for_compare */

7430

&my_charset_utf8mb4_handler,

7431

&my_collation_any_uca_handler,

7432

PAD_SPACE};

7433

7434

CHARSET_INFO my_charset_utf8mb4_spanish2_uca_ci = {

238,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7439

MY_UTF8MB4, /* csname */

7440

MY_UTF8MB4 "_spanish2_ci", /* m_coll_name */

7441

"UTF-8 Unicode", /* comment */

7442

spanish2, /* tailoring */

7443

nullptr, /* coll_param */

7444

ctype_utf8, /* ctype */

7445

nullptr, /* to_lower */

7446

nullptr, /* to_upper */

7447

nullptr, /* sort_order */

7448

nullptr, /* uca */

7449

nullptr, /* tab_to_uni */

7450

nullptr, /* tab_from_uni */

7451

&my_unicase_default, /* caseinfo */

7452

nullptr, /* state_map */

7453

nullptr, /* ident_map */

7454

8, /* strxfrm_multiply */

7455

1, /* caseup_multiply */

7456

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7461

0xFFFF, /* max_sort_char */

7462

' ', /* pad char */

7463

false, /* escape_with_backslash_is_dangerous */

7464

1, /* levels_for_compare */

7465

&my_charset_utf8mb4_handler,

7466

&my_collation_any_uca_handler,

7467

PAD_SPACE};

7468

7469

CHARSET_INFO my_charset_utf8mb4_roman_uca_ci = {

239,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7474

MY_UTF8MB4, /* csname */

7475

MY_UTF8MB4 "_roman_ci", /* m_coll_name */

7476

"UTF-8 Unicode", /* comment */

7477

roman, /* tailoring */

7478

nullptr, /* coll_param */

7479

ctype_utf8, /* ctype */

7480

nullptr, /* to_lower */

7481

nullptr, /* to_upper */

7482

nullptr, /* sort_order */

7483

nullptr, /* uca */

7484

nullptr, /* tab_to_uni */

7485

nullptr, /* tab_from_uni */

7486

&my_unicase_default, /* caseinfo */

7487

nullptr, /* state_map */

7488

nullptr, /* ident_map */

7489

8, /* strxfrm_multiply */

7490

1, /* caseup_multiply */

7491

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7496

0xFFFF, /* max_sort_char */

7497

' ', /* pad char */

7498

false, /* escape_with_backslash_is_dangerous */

7499

1, /* levels_for_compare */

7500

&my_charset_utf8mb4_handler,

7501

&my_collation_any_uca_handler,

7502

PAD_SPACE};

7503

7504

CHARSET_INFO my_charset_utf8mb4_persian_uca_ci = {

240,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7509

MY_UTF8MB4, /* csname */

7510

MY_UTF8MB4 "_persian_ci", /* m_coll_name */

7511

"UTF-8 Unicode", /* comment */

7512

persian, /* tailoring */

7513

nullptr, /* coll_param */

7514

ctype_utf8, /* ctype */

7515

nullptr, /* to_lower */

7516

nullptr, /* to_upper */

7517

nullptr, /* sort_order */

7518

nullptr, /* uca */

7519

nullptr, /* tab_to_uni */

7520

nullptr, /* tab_from_uni */

7521

&my_unicase_default, /* caseinfo */

7522

nullptr, /* state_map */

7523

nullptr, /* ident_map */

7524

8, /* strxfrm_multiply */

7525

1, /* caseup_multiply */

7526

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7531

0xFFFF, /* max_sort_char */

7532

' ', /* pad char */

7533

false, /* escape_with_backslash_is_dangerous */

7534

1, /* levels_for_compare */

7535

&my_charset_utf8mb4_handler,

7536

&my_collation_any_uca_handler,

7537

PAD_SPACE};

7538

7539

CHARSET_INFO my_charset_utf8mb4_esperanto_uca_ci = {

241,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7544

MY_UTF8MB4, /* csname */

7545

MY_UTF8MB4 "_esperanto_ci", /* m_coll_name */

7546

"UTF-8 Unicode", /* comment */

7547

esperanto, /* tailoring */

7548

nullptr, /* coll_param */

7549

ctype_utf8, /* ctype */

7550

nullptr, /* to_lower */

7551

nullptr, /* to_upper */

7552

nullptr, /* sort_order */

7553

nullptr, /* uca */

7554

nullptr, /* tab_to_uni */

7555

nullptr, /* tab_from_uni */

7556

&my_unicase_default, /* caseinfo */

7557

nullptr, /* state_map */

7558

nullptr, /* ident_map */

7559

8, /* strxfrm_multiply */

7560

1, /* caseup_multiply */

7561

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7566

0xFFFF, /* max_sort_char */

7567

' ', /* pad char */

7568

false, /* escape_with_backslash_is_dangerous */

7569

1, /* levels_for_compare */

7570

&my_charset_utf8mb4_handler,

7571

&my_collation_any_uca_handler,

7572

PAD_SPACE};

7573

7574

CHARSET_INFO my_charset_utf8mb4_hungarian_uca_ci = {

242,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7579

MY_UTF8MB4, /* csname */

7580

MY_UTF8MB4 "_hungarian_ci", /* m_coll_name */

7581

"UTF-8 Unicode", /* comment */

7582

hungarian, /* tailoring */

7583

nullptr, /* coll_param */

7584

ctype_utf8, /* ctype */

7585

nullptr, /* to_lower */

7586

nullptr, /* to_upper */

7587

nullptr, /* sort_order */

7588

nullptr, /* uca */

7589

nullptr, /* tab_to_uni */

7590

nullptr, /* tab_from_uni */

7591

&my_unicase_default, /* caseinfo */

7592

nullptr, /* state_map */

7593

nullptr, /* ident_map */

7594

8, /* strxfrm_multiply */

7595

1, /* caseup_multiply */

7596

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7601

0xFFFF, /* max_sort_char */

7602

' ', /* pad char */

7603

false, /* escape_with_backslash_is_dangerous */

7604

1, /* levels_for_compare */

7605

&my_charset_utf8mb4_handler,

7606

&my_collation_any_uca_handler,

7607

PAD_SPACE};

7608

7609

CHARSET_INFO my_charset_utf8mb4_sinhala_uca_ci = {

243,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7614

MY_UTF8MB4, /* csname */

7615

MY_UTF8MB4 "_sinhala_ci", /* m_coll_name */

7616

"UTF-8 Unicode", /* comment */

7617

sinhala, /* tailoring */

7618

nullptr, /* coll_param */

7619

ctype_utf8, /* ctype */

7620

nullptr, /* to_lower */

7621

nullptr, /* to_upper */

7622

nullptr, /* sort_order */

7623

nullptr, /* uca */

7624

nullptr, /* tab_to_uni */

7625

nullptr, /* tab_from_uni */

7626

&my_unicase_default, /* caseinfo */

7627

nullptr, /* state_map */

7628

nullptr, /* ident_map */

7629

8, /* strxfrm_multiply */

7630

1, /* caseup_multiply */

7631

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7636

0xFFFF, /* max_sort_char */

7637

' ', /* pad char */

7638

false, /* escape_with_backslash_is_dangerous */

7639

1, /* levels_for_compare */

7640

&my_charset_utf8mb4_handler,

7641

&my_collation_any_uca_handler,

7642

PAD_SPACE};

7643

7644

CHARSET_INFO my_charset_utf8mb4_german2_uca_ci = {

244,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7649

MY_UTF8MB4, /* csname */

7650

MY_UTF8MB4 "_german2_ci", /* m_coll_name */

7651

"UTF-8 Unicode", /* comment */

7652

german2, /* tailoring */

7653

nullptr, /* coll_param */

7654

ctype_utf8, /* ctype */

7655

nullptr, /* to_lower */

7656

nullptr, /* to_upper */

7657

nullptr, /* sort_order */

7658

nullptr, /* uca */

7659

nullptr, /* tab_to_uni */

7660

nullptr, /* tab_from_uni */

7661

&my_unicase_default, /* caseinfo */

7662

nullptr, /* state_map */

7663

nullptr, /* ident_map */

7664

8, /* strxfrm_multiply */

7665

1, /* caseup_multiply */

7666

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7671

0xFFFF, /* max_sort_char */

7672

' ', /* pad char */

7673

false, /* escape_with_backslash_is_dangerous */

7674

1, /* levels_for_compare */

7675

&my_charset_utf8mb4_handler,

7676

&my_collation_any_uca_handler,

7677

PAD_SPACE};

7678

7679

CHARSET_INFO my_charset_utf8mb4_croatian_uca_ci = {

245,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7684

MY_UTF8MB4, /* csname */

7685

MY_UTF8MB4 "_croatian_ci", /* m_coll_name */

7686

"UTF-8 Unicode", /* comment */

7687

croatian, /* tailoring */

7688

nullptr, /* coll_param */

7689

ctype_utf8, /* ctype */

7690

nullptr, /* to_lower */

7691

nullptr, /* to_upper */

7692

nullptr, /* sort_order */

7693

nullptr, /* uca */

7694

nullptr, /* tab_to_uni */

7695

nullptr, /* tab_from_uni */

7696

&my_unicase_default, /* caseinfo */

7697

nullptr, /* state_map */

7698

nullptr, /* ident_map */

7699

8, /* strxfrm_multiply */

7700

1, /* caseup_multiply */

7701

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7706

0xFFFF, /* max_sort_char */

7707

' ', /* pad char */

7708

false, /* escape_with_backslash_is_dangerous */

7709

1, /* levels_for_compare */

7710

&my_charset_utf8mb4_handler,

7711

&my_collation_any_uca_handler,

7712

PAD_SPACE};

7713

7714

CHARSET_INFO my_charset_utf8mb4_unicode_520_ci = {

246,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* flags */

7719

MY_UTF8MB4, /* csname */

7720

MY_UTF8MB4 "_unicode_520_ci", /* m_coll_name */

7721

"UTF-8 Unicode", /* comment */

7722

"", /* tailoring */

7723

nullptr, /* coll_param */

7724

ctype_utf8, /* ctype */

7725

nullptr, /* to_lower */

7726

nullptr, /* to_upper */

7727

nullptr, /* sort_order */

7728

&my_uca_v520, /* uca */

7729

nullptr, /* tab_to_uni */

7730

nullptr, /* tab_from_uni */

7731

&my_unicase_unicode520, /* caseinfo */

7732

nullptr, /* state_map */

7733

nullptr, /* ident_map */

7734

8, /* strxfrm_multiply */

7735

1, /* caseup_multiply */

7736

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7741

0x10FFFF, /* max_sort_char */

7742

' ', /* pad char */

7743

false, /* escape_with_backslash_is_dangerous */

7744

1, /* levels_for_compare */

7745

&my_charset_utf8mb4_handler,

7746

&my_collation_any_uca_handler,

7747

PAD_SPACE};

7748

7749

CHARSET_INFO my_charset_utf8mb4_vietnamese_ci = {

247,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

7754

MY_UTF8MB4, /* csname */

7755

MY_UTF8MB4 "_vietnamese_ci", /* m_coll_name */

7756

"UTF-8 Unicode", /* comment */

7757

vietnamese, /* tailoring */

7758

nullptr, /* coll_param */

7759

ctype_utf8, /* ctype */

7760

nullptr, /* to_lower */

7761

nullptr, /* to_upper */

7762

nullptr, /* sort_order */

7763

nullptr, /* uca */

7764

nullptr, /* tab_to_uni */

7765

nullptr, /* tab_from_uni */

7766

&my_unicase_default, /* caseinfo */

7767

nullptr, /* state_map */

7768

nullptr, /* ident_map */

7769

8, /* strxfrm_multiply */

7770

1, /* caseup_multiply */

7771

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7776

0xFFFF, /* max_sort_char */

7777

' ', /* pad char */

7778

false, /* escape_with_backslash_is_dangerous */

7779

1, /* levels_for_compare */

7780

&my_charset_utf8mb4_handler,

7781

&my_collation_any_uca_handler,

7782

PAD_SPACE};

7783

7784

MY_COLLATION_HANDLER my_collation_utf32_uca_handler = {

7785

my_coll_init_uca, /* init */

7786

my_coll_uninit_uca,

7787

my_strnncoll_any_uca,

7788

my_strnncollsp_any_uca,

7789

my_strnxfrm_any_uca,

7790

my_strnxfrmlen_simple,

7791

my_like_range_generic,

my_wildcmp_uca,

nullptr,

my_instr_mb,

my_hash_sort_any_uca,

7796

my_propagate_complex};

7797

7798

extern MY_CHARSET_HANDLER my_charset_utf32_handler;

7799

7800

#define MY_CS_UTF32_UCA_FLAGS \

7801

(MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | \

7802

MY_CS_UNICODE_SUPPLEMENT | MY_CS_NONASCII)

7803

7804

CHARSET_INFO my_charset_utf32_unicode_ci = {

160,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

7809

"utf32", /* csname */

7810

"utf32_unicode_ci", /* m_coll_name */

7811

"", /* comment */

7812

"", /* tailoring */

7813

nullptr, /* coll_param */

7814

nullptr, /* ctype */

7815

nullptr, /* to_lower */

7816

nullptr, /* to_upper */

7817

nullptr, /* sort_order */

7818

nullptr, /* uca */

7819

nullptr, /* tab_to_uni */

7820

nullptr, /* tab_from_uni */

7821

&my_unicase_default, /* caseinfo */

7822

nullptr, /* state_map */

7823

nullptr, /* ident_map */

7824

8, /* strxfrm_multiply */

7825

1, /* caseup_multiply */

7826

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7831

0xFFFF, /* max_sort_char */

7832

' ', /* pad char */

7833

false, /* escape_with_backslash_is_dangerous */

7834

1, /* levels_for_compare */

7835

&my_charset_utf32_handler,

7836

&my_collation_utf32_uca_handler,

7837

PAD_SPACE};

7838

7839

CHARSET_INFO my_charset_utf32_icelandic_uca_ci = {

161,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

7844

"utf32", /* csname */

7845

"utf32_icelandic_ci", /* m_coll_name */

7846

"", /* comment */

7847

icelandic, /* tailoring */

7848

nullptr, /* coll_param */

7849

nullptr, /* ctype */

7850

nullptr, /* to_lower */

7851

nullptr, /* to_upper */

7852

nullptr, /* sort_order */

7853

nullptr, /* uca */

7854

nullptr, /* tab_to_uni */

7855

nullptr, /* tab_from_uni */

7856

&my_unicase_default, /* caseinfo */

7857

nullptr, /* state_map */

7858

nullptr, /* ident_map */

7859

8, /* strxfrm_multiply */

7860

1, /* caseup_multiply */

7861

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7866

0xFFFF, /* max_sort_char */

7867

' ', /* pad char */

7868

false, /* escape_with_backslash_is_dangerous */

7869

1, /* levels_for_compare */

7870

&my_charset_utf32_handler,

7871

&my_collation_utf32_uca_handler,

7872

PAD_SPACE};

7873

7874

CHARSET_INFO my_charset_utf32_latvian_uca_ci = {

162,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

7879

"utf32", /* csname */

7880

"utf32_latvian_ci", /* m_coll_name */

7881

"", /* comment */

7882

latvian, /* tailoring */

7883

nullptr, /* coll_param */

7884

nullptr, /* ctype */

7885

nullptr, /* to_lower */

7886

nullptr, /* to_upper */

7887

nullptr, /* sort_order */

7888

nullptr, /* uca */

7889

nullptr, /* tab_to_uni */

7890

nullptr, /* tab_from_uni */

7891

&my_unicase_default, /* caseinfo */

7892

nullptr, /* state_map */

7893

nullptr, /* ident_map */

7894

8, /* strxfrm_multiply */

7895

1, /* caseup_multiply */

7896

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7901

0xFFFF, /* max_sort_char */

7902

' ', /* pad char */

7903

false, /* escape_with_backslash_is_dangerous */

7904

1, /* levels_for_compare */

7905

&my_charset_utf32_handler,

7906

&my_collation_utf32_uca_handler,

7907

PAD_SPACE};

7908

7909

CHARSET_INFO my_charset_utf32_romanian_uca_ci = {

163,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

7914

"utf32", /* csname */

7915

"utf32_romanian_ci", /* m_coll_name */

7916

"", /* comment */

7917

romanian, /* tailoring */

7918

nullptr, /* coll_param */

7919

nullptr, /* ctype */

7920

nullptr, /* to_lower */

7921

nullptr, /* to_upper */

7922

nullptr, /* sort_order */

7923

nullptr, /* uca */

7924

nullptr, /* tab_to_uni */

7925

nullptr, /* tab_from_uni */

7926

&my_unicase_default, /* caseinfo */

7927

nullptr, /* state_map */

7928

nullptr, /* ident_map */

7929

8, /* strxfrm_multiply */

7930

1, /* caseup_multiply */

7931

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7936

0xFFFF, /* max_sort_char */

7937

' ', /* pad char */

7938

false, /* escape_with_backslash_is_dangerous */

7939

1, /* levels_for_compare */

7940

&my_charset_utf32_handler,

7941

&my_collation_utf32_uca_handler,

7942

PAD_SPACE};

7943

7944

CHARSET_INFO my_charset_utf32_slovenian_uca_ci = {

164,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

7949

"utf32", /* csname */

7950

"utf32_slovenian_ci", /* m_coll_name */

7951

"", /* comment */

7952

slovenian, /* tailoring */

7953

nullptr, /* coll_param */

7954

nullptr, /* ctype */

7955

nullptr, /* to_lower */

7956

nullptr, /* to_upper */

7957

nullptr, /* sort_order */

7958

nullptr, /* uca */

7959

nullptr, /* tab_to_uni */

7960

nullptr, /* tab_from_uni */

7961

&my_unicase_default, /* caseinfo */

7962

nullptr, /* state_map */

7963

nullptr, /* ident_map */

7964

8, /* strxfrm_multiply */

7965

1, /* caseup_multiply */

7966

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

7971

0xFFFF, /* max_sort_char */

7972

' ', /* pad char */

7973

false, /* escape_with_backslash_is_dangerous */

7974

1, /* levels_for_compare */

7975

&my_charset_utf32_handler,

7976

&my_collation_utf32_uca_handler,

7977

PAD_SPACE};

7978

7979

CHARSET_INFO my_charset_utf32_polish_uca_ci = {

165,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

7984

"utf32", /* csname */

7985

"utf32_polish_ci", /* m_coll_name */

7986

"", /* comment */

7987

polish, /* tailoring */

7988

nullptr, /* coll_param */

7989

nullptr, /* ctype */

7990

nullptr, /* to_lower */

7991

nullptr, /* to_upper */

7992

nullptr, /* sort_order */

7993

nullptr, /* uca */

7994

nullptr, /* tab_to_uni */

7995

nullptr, /* tab_from_uni */

7996

&my_unicase_default, /* caseinfo */

7997

nullptr, /* state_map */

7998

nullptr, /* ident_map */

7999

8, /* strxfrm_multiply */

8000

1, /* caseup_multiply */

8001

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8006

0xFFFF, /* max_sort_char */

8007

' ', /* pad char */

8008

false, /* escape_with_backslash_is_dangerous */

8009

1, /* levels_for_compare */

8010

&my_charset_utf32_handler,

8011

&my_collation_utf32_uca_handler,

8012

PAD_SPACE};

8013

8014

CHARSET_INFO my_charset_utf32_estonian_uca_ci = {

166,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8019

"utf32", /* csname */

8020

"utf32_estonian_ci", /* m_coll_name */

8021

"", /* comment */

8022

estonian, /* tailoring */

8023

nullptr, /* coll_param */

8024

nullptr, /* ctype */

8025

nullptr, /* to_lower */

8026

nullptr, /* to_upper */

8027

nullptr, /* sort_order */

8028

nullptr, /* uca */

8029

nullptr, /* tab_to_uni */

8030

nullptr, /* tab_from_uni */

8031

&my_unicase_default, /* caseinfo */

8032

nullptr, /* state_map */

8033

nullptr, /* ident_map */

8034

8, /* strxfrm_multiply */

8035

1, /* caseup_multiply */

8036

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8041

0xFFFF, /* max_sort_char */

8042

' ', /* pad char */

8043

false, /* escape_with_backslash_is_dangerous */

8044

1, /* levels_for_compare */

8045

&my_charset_utf32_handler,

8046

&my_collation_utf32_uca_handler,

8047

PAD_SPACE};

8048

8049

CHARSET_INFO my_charset_utf32_spanish_uca_ci = {

167,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8054

"utf32", /* csname */

8055

"utf32_spanish_ci", /* m_coll_name */

8056

"", /* comment */

8057

spanish, /* tailoring */

8058

nullptr, /* coll_param */

8059

nullptr, /* ctype */

8060

nullptr, /* to_lower */

8061

nullptr, /* to_upper */

8062

nullptr, /* sort_order */

8063

nullptr, /* uca */

8064

nullptr, /* tab_to_uni */

8065

nullptr, /* tab_from_uni */

8066

&my_unicase_default, /* caseinfo */

8067

nullptr, /* state_map */

8068

nullptr, /* ident_map */

8069

8, /* strxfrm_multiply */

8070

1, /* caseup_multiply */

8071

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8076

0xFFFF, /* max_sort_char */

8077

' ', /* pad char */

8078

false, /* escape_with_backslash_is_dangerous */

8079

1, /* levels_for_compare */

8080

&my_charset_utf32_handler,

8081

&my_collation_utf32_uca_handler,

8082

PAD_SPACE};

8083

8084

CHARSET_INFO my_charset_utf32_swedish_uca_ci = {

168,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8089

"utf32", /* csname */

8090

"utf32_swedish_ci", /* m_coll_name */

8091

"", /* comment */

8092

swedish, /* tailoring */

8093

nullptr, /* coll_param */

8094

nullptr, /* ctype */

8095

nullptr, /* to_lower */

8096

nullptr, /* to_upper */

8097

nullptr, /* sort_order */

8098

nullptr, /* uca */

8099

nullptr, /* tab_to_uni */

8100

nullptr, /* tab_from_uni */

8101

&my_unicase_default, /* caseinfo */

8102

nullptr, /* state_map */

8103

nullptr, /* ident_map */

8104

8, /* strxfrm_multiply */

8105

1, /* caseup_multiply */

8106

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8111

0xFFFF, /* max_sort_char */

8112

' ', /* pad char */

8113

false, /* escape_with_backslash_is_dangerous */

8114

1, /* levels_for_compare */

8115

&my_charset_utf32_handler,

8116

&my_collation_utf32_uca_handler,

8117

PAD_SPACE};

8118

8119

CHARSET_INFO my_charset_utf32_turkish_uca_ci = {

169,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8124

"utf32", /* csname */

8125

"utf32_turkish_ci", /* m_coll_name */

8126

"", /* comment */

8127

turkish, /* tailoring */

8128

nullptr, /* coll_param */

8129

nullptr, /* ctype */

8130

nullptr, /* to_lower */

8131

nullptr, /* to_upper */

8132

nullptr, /* sort_order */

8133

nullptr, /* uca */

8134

nullptr, /* tab_to_uni */

8135

nullptr, /* tab_from_uni */

8136

&my_unicase_turkish, /* caseinfo */

8137

nullptr, /* state_map */

8138

nullptr, /* ident_map */

8139

8, /* strxfrm_multiply */

8140

1, /* caseup_multiply */

8141

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8146

0xFFFF, /* max_sort_char */

8147

' ', /* pad char */

8148

false, /* escape_with_backslash_is_dangerous */

8149

1, /* levels_for_compare */

8150

&my_charset_utf32_handler,

8151

&my_collation_utf32_uca_handler,

8152

PAD_SPACE};

8153

8154

CHARSET_INFO my_charset_utf32_czech_uca_ci = {

170,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8159

"utf32", /* csname */

8160

"utf32_czech_ci", /* m_coll_name */

8161

"", /* comment */

8162

czech, /* tailoring */

8163

nullptr, /* coll_param */

8164

nullptr, /* ctype */

8165

nullptr, /* to_lower */

8166

nullptr, /* to_upper */

8167

nullptr, /* sort_order */

8168

nullptr, /* uca */

8169

nullptr, /* tab_to_uni */

8170

nullptr, /* tab_from_uni */

8171

&my_unicase_default, /* caseinfo */

8172

nullptr, /* state_map */

8173

nullptr, /* ident_map */

8174

8, /* strxfrm_multiply */

8175

1, /* caseup_multiply */

8176

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8181

0xFFFF, /* max_sort_char */

8182

' ', /* pad char */

8183

false, /* escape_with_backslash_is_dangerous */

8184

1, /* levels_for_compare */

8185

&my_charset_utf32_handler,

8186

&my_collation_utf32_uca_handler,

8187

PAD_SPACE};

8188

8189

CHARSET_INFO my_charset_utf32_danish_uca_ci = {

171,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8194

"utf32", /* csname */

8195

"utf32_danish_ci", /* m_coll_name */

8196

"", /* comment */

8197

danish, /* tailoring */

8198

nullptr, /* coll_param */

8199

nullptr, /* ctype */

8200

nullptr, /* to_lower */

8201

nullptr, /* to_upper */

8202

nullptr, /* sort_order */

8203

nullptr, /* uca */

8204

nullptr, /* tab_to_uni */

8205

nullptr, /* tab_from_uni */

8206

&my_unicase_default, /* caseinfo */

8207

nullptr, /* state_map */

8208

nullptr, /* ident_map */

8209

8, /* strxfrm_multiply */

8210

1, /* caseup_multiply */

8211

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8216

0xFFFF, /* max_sort_char */

8217

' ', /* pad char */

8218

false, /* escape_with_backslash_is_dangerous */

8219

1, /* levels_for_compare */

8220

&my_charset_utf32_handler,

8221

&my_collation_utf32_uca_handler,

8222

PAD_SPACE};

8223

8224

CHARSET_INFO my_charset_utf32_lithuanian_uca_ci = {

172,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8229

"utf32", /* csname */

8230

"utf32_lithuanian_ci", /* m_coll_name */

8231

"", /* comment */

8232

lithuanian, /* tailoring */

8233

nullptr, /* coll_param */

8234

nullptr, /* ctype */

8235

nullptr, /* to_lower */

8236

nullptr, /* to_upper */

8237

nullptr, /* sort_order */

8238

nullptr, /* uca */

8239

nullptr, /* tab_to_uni */

8240

nullptr, /* tab_from_uni */

8241

&my_unicase_default, /* caseinfo */

8242

nullptr, /* state_map */

8243

nullptr, /* ident_map */

8244

8, /* strxfrm_multiply */

8245

1, /* caseup_multiply */

8246

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8251

0xFFFF, /* max_sort_char */

8252

' ', /* pad char */

8253

false, /* escape_with_backslash_is_dangerous */

8254

1, /* levels_for_compare */

8255

&my_charset_utf32_handler,

8256

&my_collation_utf32_uca_handler,

8257

PAD_SPACE};

8258

8259

CHARSET_INFO my_charset_utf32_slovak_uca_ci = {

173,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8264

"utf32", /* csname */

8265

"utf32_slovak_ci", /* m_coll_name */

8266

"", /* comment */

8267

slovak, /* tailoring */

8268

nullptr, /* coll_param */

8269

nullptr, /* ctype */

8270

nullptr, /* to_lower */

8271

nullptr, /* to_upper */

8272

nullptr, /* sort_order */

8273

nullptr, /* uca */

8274

nullptr, /* tab_to_uni */

8275

nullptr, /* tab_from_uni */

8276

&my_unicase_default, /* caseinfo */

8277

nullptr, /* state_map */

8278

nullptr, /* ident_map */

8279

8, /* strxfrm_multiply */

8280

1, /* caseup_multiply */

8281

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8286

0xFFFF, /* max_sort_char */

8287

' ', /* pad char */

8288

false, /* escape_with_backslash_is_dangerous */

8289

1, /* levels_for_compare */

8290

&my_charset_utf32_handler,

8291

&my_collation_utf32_uca_handler,

8292

PAD_SPACE};

8293

8294

CHARSET_INFO my_charset_utf32_spanish2_uca_ci = {

174,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8299

"utf32", /* csname */

8300

"utf32_spanish2_ci", /* m_coll_name */

8301

"", /* comment */

8302

spanish2, /* tailoring */

8303

nullptr, /* coll_param */

8304

nullptr, /* ctype */

8305

nullptr, /* to_lower */

8306

nullptr, /* to_upper */

8307

nullptr, /* sort_order */

8308

nullptr, /* uca */

8309

nullptr, /* tab_to_uni */

8310

nullptr, /* tab_from_uni */

8311

&my_unicase_default, /* caseinfo */

8312

nullptr, /* state_map */

8313

nullptr, /* ident_map */

8314

8, /* strxfrm_multiply */

8315

1, /* caseup_multiply */

8316

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8321

0xFFFF, /* max_sort_char */

8322

' ', /* pad char */

8323

false, /* escape_with_backslash_is_dangerous */

8324

1, /* levels_for_compare */

8325

&my_charset_utf32_handler,

8326

&my_collation_utf32_uca_handler,

8327

PAD_SPACE};

8328

8329

CHARSET_INFO my_charset_utf32_roman_uca_ci = {

175,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8334

"utf32", /* csname */

8335

"utf32_roman_ci", /* m_coll_name */

8336

"", /* comment */

8337

roman, /* tailoring */

8338

nullptr, /* coll_param */

8339

nullptr, /* ctype */

8340

nullptr, /* to_lower */

8341

nullptr, /* to_upper */

8342

nullptr, /* sort_order */

8343

nullptr, /* uca */

8344

nullptr, /* tab_to_uni */

8345

nullptr, /* tab_from_uni */

8346

&my_unicase_default, /* caseinfo */

8347

nullptr, /* state_map */

8348

nullptr, /* ident_map */

8349

8, /* strxfrm_multiply */

8350

1, /* caseup_multiply */

8351

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8356

0xFFFF, /* max_sort_char */

8357

' ', /* pad char */

8358

false, /* escape_with_backslash_is_dangerous */

8359

1, /* levels_for_compare */

8360

&my_charset_utf32_handler,

8361

&my_collation_utf32_uca_handler,

8362

PAD_SPACE};

8363

8364

CHARSET_INFO my_charset_utf32_persian_uca_ci = {

176,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8369

"utf32", /* csname */

8370

"utf32_persian_ci", /* m_coll_name */

8371

"", /* comment */

8372

persian, /* tailoring */

8373

nullptr, /* coll_param */

8374

nullptr, /* ctype */

8375

nullptr, /* to_lower */

8376

nullptr, /* to_upper */

8377

nullptr, /* sort_order */

8378

nullptr, /* uca */

8379

nullptr, /* tab_to_uni */

8380

nullptr, /* tab_from_uni */

8381

&my_unicase_default, /* caseinfo */

8382

nullptr, /* state_map */

8383

nullptr, /* ident_map */

8384

8, /* strxfrm_multiply */

8385

1, /* caseup_multiply */

8386

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8391

0xFFFF, /* max_sort_char */

8392

' ', /* pad char */

8393

false, /* escape_with_backslash_is_dangerous */

8394

1, /* levels_for_compare */

8395

&my_charset_utf32_handler,

8396

&my_collation_utf32_uca_handler,

8397

PAD_SPACE};

8398

8399

CHARSET_INFO my_charset_utf32_esperanto_uca_ci = {

177,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8404

"utf32", /* csname */

8405

"utf32_esperanto_ci", /* m_coll_name */

8406

"", /* comment */

8407

esperanto, /* tailoring */

8408

nullptr, /* coll_param */

8409

nullptr, /* ctype */

8410

nullptr, /* to_lower */

8411

nullptr, /* to_upper */

8412

nullptr, /* sort_order */

8413

nullptr, /* uca */

8414

nullptr, /* tab_to_uni */

8415

nullptr, /* tab_from_uni */

8416

&my_unicase_default, /* caseinfo */

8417

nullptr, /* state_map */

8418

nullptr, /* ident_map */

8419

8, /* strxfrm_multiply */

8420

1, /* caseup_multiply */

8421

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8426

0xFFFF, /* max_sort_char */

8427

' ', /* pad char */

8428

false, /* escape_with_backslash_is_dangerous */

8429

1, /* levels_for_compare */

8430

&my_charset_utf32_handler,

8431

&my_collation_utf32_uca_handler,

8432

PAD_SPACE};

8433

8434

CHARSET_INFO my_charset_utf32_hungarian_uca_ci = {

178,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8439

"utf32", /* csname */

8440

"utf32_hungarian_ci", /* m_coll_name */

8441

"", /* comment */

8442

hungarian, /* tailoring */

8443

nullptr, /* coll_param */

8444

nullptr, /* ctype */

8445

nullptr, /* to_lower */

8446

nullptr, /* to_upper */

8447

nullptr, /* sort_order */

8448

nullptr, /* uca */

8449

nullptr, /* tab_to_uni */

8450

nullptr, /* tab_from_uni */

8451

&my_unicase_default, /* caseinfo */

8452

nullptr, /* state_map */

8453

nullptr, /* ident_map */

8454

8, /* strxfrm_multiply */

8455

1, /* caseup_multiply */

8456

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8461

0xFFFF, /* max_sort_char */

8462

' ', /* pad char */

8463

false, /* escape_with_backslash_is_dangerous */

8464

1, /* levels_for_compare */

8465

&my_charset_utf32_handler,

8466

&my_collation_utf32_uca_handler,

8467

PAD_SPACE};

8468

8469

CHARSET_INFO my_charset_utf32_sinhala_uca_ci = {

179,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8474

"utf32", /* csname */

8475

"utf32_sinhala_ci", /* m_coll_name */

8476

"", /* comment */

8477

sinhala, /* tailoring */

8478

nullptr, /* coll_param */

8479

nullptr, /* ctype */

8480

nullptr, /* to_lower */

8481

nullptr, /* to_upper */

8482

nullptr, /* sort_order */

8483

nullptr, /* uca */

8484

nullptr, /* tab_to_uni */

8485

nullptr, /* tab_from_uni */

8486

&my_unicase_default, /* caseinfo */

8487

nullptr, /* state_map */

8488

nullptr, /* ident_map */

8489

8, /* strxfrm_multiply */

8490

1, /* caseup_multiply */

8491

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8496

0xFFFF, /* max_sort_char */

8497

' ', /* pad char */

8498

false, /* escape_with_backslash_is_dangerous */

8499

1, /* levels_for_compare */

8500

&my_charset_utf32_handler,

8501

&my_collation_utf32_uca_handler,

8502

PAD_SPACE};

8503

8504

CHARSET_INFO my_charset_utf32_german2_uca_ci = {

180,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8509

"utf32", /* csname */

8510

"utf32_german2_ci", /* m_coll_name */

8511

"", /* comment */

8512

german2, /* tailoring */

8513

nullptr, /* coll_param */

8514

nullptr, /* ctype */

8515

nullptr, /* to_lower */

8516

nullptr, /* to_upper */

8517

nullptr, /* sort_order */

8518

nullptr, /* uca */

8519

nullptr, /* tab_to_uni */

8520

nullptr, /* tab_from_uni */

8521

&my_unicase_default, /* caseinfo */

8522

nullptr, /* state_map */

8523

nullptr, /* ident_map */

8524

8, /* strxfrm_multiply */

8525

1, /* caseup_multiply */

8526

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8531

0xFFFF, /* max_sort_char */

8532

' ', /* pad char */

8533

false, /* escape_with_backslash_is_dangerous */

8534

1, /* levels_for_compare */

8535

&my_charset_utf32_handler,

8536

&my_collation_utf32_uca_handler,

8537

PAD_SPACE};

8538

8539

CHARSET_INFO my_charset_utf32_croatian_uca_ci = {

181,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8544

"utf32", /* csname */

8545

"utf32_croatian_ci", /* m_coll_name */

8546

"", /* comment */

8547

croatian, /* tailoring */

8548

nullptr, /* coll_param */

8549

nullptr, /* ctype */

8550

nullptr, /* to_lower */

8551

nullptr, /* to_upper */

8552

nullptr, /* sort_order */

8553

nullptr, /* uca */

8554

nullptr, /* tab_to_uni */

8555

nullptr, /* tab_from_uni */

8556

&my_unicase_default, /* caseinfo */

8557

nullptr, /* state_map */

8558

nullptr, /* ident_map */

8559

8, /* strxfrm_multiply */

8560

1, /* caseup_multiply */

8561

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8566

0xFFFF, /* max_sort_char */

8567

' ', /* pad char */

8568

false, /* escape_with_backslash_is_dangerous */

8569

1, /* levels_for_compare */

8570

&my_charset_utf32_handler,

8571

&my_collation_utf32_uca_handler,

8572

PAD_SPACE};

8573

8574

CHARSET_INFO my_charset_utf32_unicode_520_ci = {

182,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* stat e */

8579

"utf32", /* csname */

8580

"utf32_unicode_520_ci", /* m_coll_name */

8581

"", /* comment */

8582

"", /* tailoring */

8583

nullptr, /* coll_param */

8584

nullptr, /* ctype */

8585

nullptr, /* to_lower */

8586

nullptr, /* to_upper */

8587

nullptr, /* sort_order */

8588

&my_uca_v520, /* uca */

8589

nullptr, /* tab_to_uni */

8590

nullptr, /* tab_from_uni */

8591

&my_unicase_unicode520, /* caseinfo */

8592

nullptr, /* state_map */

8593

nullptr, /* ident_map */

8594

8, /* strxfrm_multiply */

8595

1, /* caseup_multiply */

8596

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8601

0x10FFFF, /* max_sort_char */

8602

' ', /* pad char */

8603

false, /* escape_with_backslash_is_dangerous */

8604

1, /* levels_for_compare */

8605

&my_charset_utf32_handler,

8606

&my_collation_utf32_uca_handler,

8607

PAD_SPACE};

8608

8609

CHARSET_INFO my_charset_utf32_vietnamese_ci = {

183,

0,

0, /* number */

MY_CS_UTF32_UCA_FLAGS, /* state */

8614

"utf32", /* csname */

8615

"utf32_vietnamese_ci", /* m_coll_name */

8616

"", /* comment */

8617

vietnamese, /* tailoring */

8618

nullptr, /* coll_param */

8619

nullptr, /* ctype */

8620

nullptr, /* to_lower */

8621

nullptr, /* to_upper */

8622

nullptr, /* sort_order */

8623

nullptr, /* uca */

8624

nullptr, /* tab_to_uni */

8625

nullptr, /* tab_from_uni */

8626

&my_unicase_default, /* caseinfo */

8627

nullptr, /* state_map */

8628

nullptr, /* ident_map */

8629

8, /* strxfrm_multiply */

8630

1, /* caseup_multiply */

8631

1, /* casedn_multiply */

4, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8636

0xFFFF, /* max_sort_char */

8637

' ', /* pad char */

8638

false, /* escape_with_backslash_is_dangerous */

8639

1, /* levels_for_compare */

8640

&my_charset_utf32_handler,

8641

&my_collation_utf32_uca_handler,

8642

PAD_SPACE};

8643

8644

MY_COLLATION_HANDLER my_collation_utf16_uca_handler = {

8645

my_coll_init_uca, /* init */

8646

my_coll_uninit_uca,

8647

my_strnncoll_any_uca,

8648

my_strnncollsp_any_uca,

8649

my_strnxfrm_any_uca,

8650

my_strnxfrmlen_simple,

8651

my_like_range_generic,

my_wildcmp_uca,

nullptr,

my_instr_mb,

my_hash_sort_any_uca,

8656

my_propagate_complex};

8657

8658

extern MY_CHARSET_HANDLER my_charset_utf16_handler;

8659

8660

#define MY_CS_UTF16_UCA_FLAGS \

8661

(MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII)

8662

8663

CHARSET_INFO my_charset_utf16_unicode_ci = {

101,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8668

"utf16", /* csname */

8669

"utf16_unicode_ci", /* m_coll_name */

8670

"", /* comment */

8671

"", /* tailoring */

8672

nullptr, /* coll_param */

8673

nullptr, /* ctype */

8674

nullptr, /* to_lower */

8675

nullptr, /* to_upper */

8676

nullptr, /* sort_order */

8677

nullptr, /* uca */

8678

nullptr, /* tab_to_uni */

8679

nullptr, /* tab_from_uni */

8680

&my_unicase_default, /* caseinfo */

8681

nullptr, /* state_map */

8682

nullptr, /* ident_map */

8683

8, /* strxfrm_multiply */

8684

1, /* caseup_multiply */

8685

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8690

0xFFFF, /* max_sort_char */

8691

' ', /* pad char */

8692

false, /* escape_with_backslash_is_dangerous */

8693

1, /* levels_for_compare */

8694

&my_charset_utf16_handler,

8695

&my_collation_utf16_uca_handler,

8696

PAD_SPACE};

8697

8698

CHARSET_INFO my_charset_utf16_icelandic_uca_ci = {

102,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8703

"utf16", /* csname */

8704

"utf16_icelandic_ci", /* m_coll_name */

8705

"", /* comment */

8706

icelandic, /* tailoring */

8707

nullptr, /* coll_param */

8708

nullptr, /* ctype */

8709

nullptr, /* to_lower */

8710

nullptr, /* to_upper */

8711

nullptr, /* sort_order */

8712

nullptr, /* uca */

8713

nullptr, /* tab_to_uni */

8714

nullptr, /* tab_from_uni */

8715

&my_unicase_default, /* caseinfo */

8716

nullptr, /* state_map */

8717

nullptr, /* ident_map */

8718

8, /* strxfrm_multiply */

8719

1, /* caseup_multiply */

8720

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8725

0xFFFF, /* max_sort_char */

8726

' ', /* pad char */

8727

false, /* escape_with_backslash_is_dangerous */

8728

1, /* levels_for_compare */

8729

&my_charset_utf16_handler,

8730

&my_collation_utf16_uca_handler,

8731

PAD_SPACE};

8732

8733

CHARSET_INFO my_charset_utf16_latvian_uca_ci = {

103,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8738

"utf16", /* cs name */

8739

"utf16_latvian_ci", /* m_coll_name */

8740

"", /* comment */

8741

latvian, /* tailoring */

8742

nullptr, /* coll_param */

8743

nullptr, /* ctype */

8744

nullptr, /* to_lower */

8745

nullptr, /* to_upper */

8746

nullptr, /* sort_order */

8747

nullptr, /* uca */

8748

nullptr, /* tab_to_uni */

8749

nullptr, /* tab_from_uni */

8750

&my_unicase_default, /* caseinfo */

8751

nullptr, /* state_map */

8752

nullptr, /* ident_map */

8753

8, /* strxfrm_multiply */

8754

1, /* caseup_multiply */

8755

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8760

0xFFFF, /* max_sort_char */

8761

' ', /* pad char */

8762

false, /* escape_with_backslash_is_dangerous */

8763

1, /* levels_for_compare */

8764

&my_charset_utf16_handler,

8765

&my_collation_utf16_uca_handler,

8766

PAD_SPACE};

8767

8768

CHARSET_INFO my_charset_utf16_romanian_uca_ci = {

104,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8773

"utf16", /* cs name */

8774

"utf16_romanian_ci", /* m_coll_name */

8775

"", /* comment */

8776

romanian, /* tailoring */

8777

nullptr, /* coll_param */

8778

nullptr, /* ctype */

8779

nullptr, /* to_lower */

8780

nullptr, /* to_upper */

8781

nullptr, /* sort_order */

8782

nullptr, /* uca */

8783

nullptr, /* tab_to_uni */

8784

nullptr, /* tab_from_uni */

8785

&my_unicase_default, /* caseinfo */

8786

nullptr, /* state_map */

8787

nullptr, /* ident_map */

8788

8, /* strxfrm_multiply */

8789

1, /* caseup_multiply */

8790

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8795

0xFFFF, /* max_sort_char */

8796

' ', /* pad char */

8797

false, /* escape_with_backslash_is_dangerous */

8798

1, /* levels_for_compare */

8799

&my_charset_utf16_handler,

8800

&my_collation_utf16_uca_handler,

8801

PAD_SPACE};

8802

8803

CHARSET_INFO my_charset_utf16_slovenian_uca_ci = {

105,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8808

"utf16", /* cs name */

8809

"utf16_slovenian_ci", /* m_coll_name */

8810

"", /* comment */

8811

slovenian, /* tailoring */

8812

nullptr, /* coll_param */

8813

nullptr, /* ctype */

8814

nullptr, /* to_lower */

8815

nullptr, /* to_upper */

8816

nullptr, /* sort_order */

8817

nullptr, /* uca */

8818

nullptr, /* tab_to_uni */

8819

nullptr, /* tab_from_uni */

8820

&my_unicase_default, /* caseinfo */

8821

nullptr, /* state_map */

8822

nullptr, /* ident_map */

8823

8, /* strxfrm_multiply */

8824

1, /* caseup_multiply */

8825

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8830

0xFFFF, /* max_sort_char */

8831

' ', /* pad char */

8832

false, /* escape_with_backslash_is_dangerous */

8833

1, /* levels_for_compare */

8834

&my_charset_utf16_handler,

8835

&my_collation_utf16_uca_handler,

8836

PAD_SPACE};

8837

8838

CHARSET_INFO my_charset_utf16_polish_uca_ci = {

106,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8843

"utf16", /* cs name */

8844

"utf16_polish_ci", /* m_coll_name */

8845

"", /* comment */

8846

polish, /* tailoring */

8847

nullptr, /* coll_param */

8848

nullptr, /* ctype */

8849

nullptr, /* to_lower */

8850

nullptr, /* to_upper */

8851

nullptr, /* sort_order */

8852

nullptr, /* uca */

8853

nullptr, /* tab_to_uni */

8854

nullptr, /* tab_from_uni */

8855

&my_unicase_default, /* caseinfo */

8856

nullptr, /* state_map */

8857

nullptr, /* ident_map */

8858

8, /* strxfrm_multiply */

8859

1, /* caseup_multiply */

8860

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8865

0xFFFF, /* max_sort_char */

8866

' ', /* pad char */

8867

false, /* escape_with_backslash_is_dangerous */

8868

1, /* levels_for_compare */

8869

&my_charset_utf16_handler,

8870

&my_collation_utf16_uca_handler,

8871

PAD_SPACE};

8872

8873

CHARSET_INFO my_charset_utf16_estonian_uca_ci = {

107,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8878

"utf16", /* cs name */

8879

"utf16_estonian_ci", /* m_coll_name */

8880

"", /* comment */

8881

estonian, /* tailoring */

8882

nullptr, /* coll_param */

8883

nullptr, /* ctype */

8884

nullptr, /* to_lower */

8885

nullptr, /* to_upper */

8886

nullptr, /* sort_order */

8887

nullptr, /* uca */

8888

nullptr, /* tab_to_uni */

8889

nullptr, /* tab_from_uni */

8890

&my_unicase_default, /* caseinfo */

8891

nullptr, /* state_map */

8892

nullptr, /* ident_map */

8893

8, /* strxfrm_multiply */

8894

1, /* caseup_multiply */

8895

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8900

0xFFFF, /* max_sort_char */

8901

' ', /* pad char */

8902

false, /* escape_with_backslash_is_dangerous */

8903

1, /* levels_for_compare */

8904

&my_charset_utf16_handler,

8905

&my_collation_utf16_uca_handler,

8906

PAD_SPACE};

8907

8908

CHARSET_INFO my_charset_utf16_spanish_uca_ci = {

108,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8913

"utf16", /* cs name */

8914

"utf16_spanish_ci", /* m_coll_name */

8915

"", /* comment */

8916

spanish, /* tailoring */

8917

nullptr, /* coll_param */

8918

nullptr, /* ctype */

8919

nullptr, /* to_lower */

8920

nullptr, /* to_upper */

8921

nullptr, /* sort_order */

8922

nullptr, /* uca */

8923

nullptr, /* tab_to_uni */

8924

nullptr, /* tab_from_uni */

8925

&my_unicase_default, /* caseinfo */

8926

nullptr, /* state_map */

8927

nullptr, /* ident_map */

8928

8, /* strxfrm_multiply */

8929

1, /* caseup_multiply */

8930

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8935

0xFFFF, /* max_sort_char */

8936

' ', /* pad char */

8937

false, /* escape_with_backslash_is_dangerous */

8938

1, /* levels_for_compare */

8939

&my_charset_utf16_handler,

8940

&my_collation_utf16_uca_handler,

8941

PAD_SPACE};

8942

8943

CHARSET_INFO my_charset_utf16_swedish_uca_ci = {

109,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8948

"utf16", /* cs name */

8949

"utf16_swedish_ci", /* m_coll_name */

8950

"", /* comment */

8951

swedish, /* tailoring */

8952

nullptr, /* coll_param */

8953

nullptr, /* ctype */

8954

nullptr, /* to_lower */

8955

nullptr, /* to_upper */

8956

nullptr, /* sort_order */

8957

nullptr, /* uca */

8958

nullptr, /* tab_to_uni */

8959

nullptr, /* tab_from_uni */

8960

&my_unicase_default, /* caseinfo */

8961

nullptr, /* state_map */

8962

nullptr, /* ident_map */

8963

8, /* strxfrm_multiply */

8964

1, /* caseup_multiply */

8965

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

8970

0xFFFF, /* max_sort_char */

8971

' ', /* pad char */

8972

false, /* escape_with_backslash_is_dangerous */

8973

1, /* levels_for_compare */

8974

&my_charset_utf16_handler,

8975

&my_collation_utf16_uca_handler,

8976

PAD_SPACE};

8977

8978

CHARSET_INFO my_charset_utf16_turkish_uca_ci = {

110,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

8983

"utf16", /* cs name */

8984

"utf16_turkish_ci", /* m_coll_name */

8985

"", /* comment */

8986

turkish, /* tailoring */

8987

nullptr, /* coll_param */

8988

nullptr, /* ctype */

8989

nullptr, /* to_lower */

8990

nullptr, /* to_upper */

8991

nullptr, /* sort_order */

8992

nullptr, /* uca */

8993

nullptr, /* tab_to_uni */

8994

nullptr, /* tab_from_uni */

8995

&my_unicase_turkish, /* caseinfo */

8996

nullptr, /* state_map */

8997

nullptr, /* ident_map */

8998

8, /* strxfrm_multiply */

8999

1, /* caseup_multiply */

9000

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9005

0xFFFF, /* max_sort_char */

9006

' ', /* pad char */

9007

false, /* escape_with_backslash_is_dangerous */

9008

1, /* levels_for_compare */

9009

&my_charset_utf16_handler,

9010

&my_collation_utf16_uca_handler,

9011

PAD_SPACE};

9012

9013

CHARSET_INFO my_charset_utf16_czech_uca_ci = {

111,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9018

"utf16", /* cs name */

9019

"utf16_czech_ci", /* m_coll_name */

9020

"", /* comment */

9021

czech, /* tailoring */

9022

nullptr, /* coll_param */

9023

nullptr, /* ctype */

9024

nullptr, /* to_lower */

9025

nullptr, /* to_upper */

9026

nullptr, /* sort_order */

9027

nullptr, /* uca */

9028

nullptr, /* tab_to_uni */

9029

nullptr, /* tab_from_uni */

9030

&my_unicase_default, /* caseinfo */

9031

nullptr, /* state_map */

9032

nullptr, /* ident_map */

9033

8, /* strxfrm_multiply */

9034

1, /* caseup_multiply */

9035

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9040

0xFFFF, /* max_sort_char */

9041

' ', /* pad char */

9042

false, /* escape_with_backslash_is_dangerous */

9043

1, /* levels_for_compare */

9044

&my_charset_utf16_handler,

9045

&my_collation_utf16_uca_handler,

9046

PAD_SPACE};

9047

9048

CHARSET_INFO my_charset_utf16_danish_uca_ci = {

112,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9053

"utf16", /* cs name */

9054

"utf16_danish_ci", /* m_coll_name */

9055

"", /* comment */

9056

danish, /* tailoring */

9057

nullptr, /* coll_param */

9058

nullptr, /* ctype */

9059

nullptr, /* to_lower */

9060

nullptr, /* to_upper */

9061

nullptr, /* sort_order */

9062

nullptr, /* uca */

9063

nullptr, /* tab_to_uni */

9064

nullptr, /* tab_from_uni */

9065

&my_unicase_default, /* caseinfo */

9066

nullptr, /* state_map */

9067

nullptr, /* ident_map */

9068

8, /* strxfrm_multiply */

9069

1, /* caseup_multiply */

9070

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9075

0xFFFF, /* max_sort_char */

9076

' ', /* pad char */

9077

false, /* escape_with_backslash_is_dangerous */

9078

1, /* levels_for_compare */

9079

&my_charset_utf16_handler,

9080

&my_collation_utf16_uca_handler,

9081

PAD_SPACE};

9082

9083

CHARSET_INFO my_charset_utf16_lithuanian_uca_ci = {

113,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9088

"utf16", /* cs name */

9089

"utf16_lithuanian_ci", /* m_coll_name */

9090

"", /* comment */

9091

lithuanian, /* tailoring */

9092

nullptr, /* coll_param */

9093

nullptr, /* ctype */

9094

nullptr, /* to_lower */

9095

nullptr, /* to_upper */

9096

nullptr, /* sort_order */

9097

nullptr, /* uca */

9098

nullptr, /* tab_to_uni */

9099

nullptr, /* tab_from_uni */

9100

&my_unicase_default, /* caseinfo */

9101

nullptr, /* state_map */

9102

nullptr, /* ident_map */

9103

8, /* strxfrm_multiply */

9104

1, /* caseup_multiply */

9105

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9110

0xFFFF, /* max_sort_char */

9111

' ', /* pad char */

9112

false, /* escape_with_backslash_is_dangerous */

9113

1, /* levels_for_compare */

9114

&my_charset_utf16_handler,

9115

&my_collation_utf16_uca_handler,

9116

PAD_SPACE};

9117

9118

CHARSET_INFO my_charset_utf16_slovak_uca_ci = {

114,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9123

"utf16", /* cs name */

9124

"utf16_slovak_ci", /* m_coll_name */

9125

"", /* comment */

9126

slovak, /* tailoring */

9127

nullptr, /* coll_param */

9128

nullptr, /* ctype */

9129

nullptr, /* to_lower */

9130

nullptr, /* to_upper */

9131

nullptr, /* sort_order */

9132

nullptr, /* uca */

9133

nullptr, /* tab_to_uni */

9134

nullptr, /* tab_from_uni */

9135

&my_unicase_default, /* caseinfo */

9136

nullptr, /* state_map */

9137

nullptr, /* ident_map */

9138

8, /* strxfrm_multiply */

9139

1, /* caseup_multiply */

9140

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9145

0xFFFF, /* max_sort_char */

9146

' ', /* pad char */

9147

false, /* escape_with_backslash_is_dangerous */

9148

1, /* levels_for_compare */

9149

&my_charset_utf16_handler,

9150

&my_collation_utf16_uca_handler,

9151

PAD_SPACE};

9152

9153

CHARSET_INFO my_charset_utf16_spanish2_uca_ci = {

115,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9158

"utf16", /* cs name */

9159

"utf16_spanish2_ci", /* m_coll_name */

9160

"", /* comment */

9161

spanish2, /* tailoring */

9162

nullptr, /* coll_param */

9163

nullptr, /* ctype */

9164

nullptr, /* to_lower */

9165

nullptr, /* to_upper */

9166

nullptr, /* sort_order */

9167

nullptr, /* uca */

9168

nullptr, /* tab_to_uni */

9169

nullptr, /* tab_from_uni */

9170

&my_unicase_default, /* caseinfo */

9171

nullptr, /* state_map */

9172

nullptr, /* ident_map */

9173

8, /* strxfrm_multiply */

9174

1, /* caseup_multiply */

9175

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9180

0xFFFF, /* max_sort_char */

9181

' ', /* pad char */

9182

false, /* escape_with_backslash_is_dangerous */

9183

1, /* levels_for_compare */

9184

&my_charset_utf16_handler,

9185

&my_collation_utf16_uca_handler,

9186

PAD_SPACE};

9187

9188

CHARSET_INFO my_charset_utf16_roman_uca_ci = {

116,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9193

"utf16", /* cs name */

9194

"utf16_roman_ci", /* m_coll_name */

9195

"", /* comment */

9196

roman, /* tailoring */

9197

nullptr, /* coll_param */

9198

nullptr, /* ctype */

9199

nullptr, /* to_lower */

9200

nullptr, /* to_upper */

9201

nullptr, /* sort_order */

9202

nullptr, /* uca */

9203

nullptr, /* tab_to_uni */

9204

nullptr, /* tab_from_uni */

9205

&my_unicase_default, /* caseinfo */

9206

nullptr, /* state_map */

9207

nullptr, /* ident_map */

9208

8, /* strxfrm_multiply */

9209

1, /* caseup_multiply */

9210

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9215

0xFFFF, /* max_sort_char */

9216

' ', /* pad char */

9217

false, /* escape_with_backslash_is_dangerous */

9218

1, /* levels_for_compare */

9219

&my_charset_utf16_handler,

9220

&my_collation_utf16_uca_handler,

9221

PAD_SPACE};

9222

9223

CHARSET_INFO my_charset_utf16_persian_uca_ci = {

117,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9228

"utf16", /* cs name */

9229

"utf16_persian_ci", /* m_coll_name */

9230

"", /* comment */

9231

persian, /* tailoring */

9232

nullptr, /* coll_param */

9233

nullptr, /* ctype */

9234

nullptr, /* to_lower */

9235

nullptr, /* to_upper */

9236

nullptr, /* sort_order */

9237

nullptr, /* uca */

9238

nullptr, /* tab_to_uni */

9239

nullptr, /* tab_from_uni */

9240

&my_unicase_default, /* caseinfo */

9241

nullptr, /* state_map */

9242

nullptr, /* ident_map */

9243

8, /* strxfrm_multiply */

9244

1, /* caseup_multiply */

9245

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9250

0xFFFF, /* max_sort_char */

9251

' ', /* pad char */

9252

false, /* escape_with_backslash_is_dangerous */

9253

1, /* levels_for_compare */

9254

&my_charset_utf16_handler,

9255

&my_collation_utf16_uca_handler,

9256

PAD_SPACE};

9257

9258

CHARSET_INFO my_charset_utf16_esperanto_uca_ci = {

118,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9263

"utf16", /* cs name */

9264

"utf16_esperanto_ci", /* m_coll_name */

9265

"", /* comment */

9266

esperanto, /* tailoring */

9267

nullptr, /* coll_param */

9268

nullptr, /* ctype */

9269

nullptr, /* to_lower */

9270

nullptr, /* to_upper */

9271

nullptr, /* sort_order */

9272

nullptr, /* uca */

9273

nullptr, /* tab_to_uni */

9274

nullptr, /* tab_from_uni */

9275

&my_unicase_default, /* caseinfo */

9276

nullptr, /* state_map */

9277

nullptr, /* ident_map */

9278

8, /* strxfrm_multiply */

9279

1, /* caseup_multiply */

9280

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9285

0xFFFF, /* max_sort_char */

9286

' ', /* pad char */

9287

false, /* escape_with_backslash_is_dangerous */

9288

1, /* levels_for_compare */

9289

&my_charset_utf16_handler,

9290

&my_collation_utf16_uca_handler,

9291

PAD_SPACE};

9292

9293

CHARSET_INFO my_charset_utf16_hungarian_uca_ci = {

119,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9298

"utf16", /* cs name */

9299

"utf16_hungarian_ci", /* m_coll_name */

9300

"", /* comment */

9301

hungarian, /* tailoring */

9302

nullptr, /* coll_param */

9303

nullptr, /* ctype */

9304

nullptr, /* to_lower */

9305

nullptr, /* to_upper */

9306

nullptr, /* sort_order */

9307

nullptr, /* uca */

9308

nullptr, /* tab_to_uni */

9309

nullptr, /* tab_from_uni */

9310

&my_unicase_default, /* caseinfo */

9311

nullptr, /* state_map */

9312

nullptr, /* ident_map */

9313

8, /* strxfrm_multiply */

9314

1, /* caseup_multiply */

9315

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9320

0xFFFF, /* max_sort_char */

9321

' ', /* pad char */

9322

false, /* escape_with_backslash_is_dangerous */

9323

1, /* levels_for_compare */

9324

&my_charset_utf16_handler,

9325

&my_collation_utf16_uca_handler,

9326

PAD_SPACE};

9327

9328

CHARSET_INFO my_charset_utf16_sinhala_uca_ci = {

120,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9333

"utf16", /* cs name */

9334

"utf16_sinhala_ci", /* m_coll_name */

9335

"", /* comment */

9336

sinhala, /* tailoring */

9337

nullptr, /* coll_param */

9338

nullptr, /* ctype */

9339

nullptr, /* to_lower */

9340

nullptr, /* to_upper */

9341

nullptr, /* sort_order */

9342

nullptr, /* uca */

9343

nullptr, /* tab_to_uni */

9344

nullptr, /* tab_from_uni */

9345

&my_unicase_default, /* caseinfo */

9346

nullptr, /* state_map */

9347

nullptr, /* ident_map */

9348

8, /* strxfrm_multiply */

9349

1, /* caseup_multiply */

9350

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9355

0xFFFF, /* max_sort_char */

9356

' ', /* pad char */

9357

false, /* escape_with_backslash_is_dangerous */

9358

1, /* levels_for_compare */

9359

&my_charset_utf16_handler,

9360

&my_collation_utf16_uca_handler,

9361

PAD_SPACE};

9362

9363

CHARSET_INFO my_charset_utf16_german2_uca_ci = {

121,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9368

"utf16", /* cs name */

9369

"utf16_german2_ci", /* m_coll_name */

9370

"", /* comment */

9371

german2, /* tailoring */

9372

nullptr, /* coll_param */

9373

nullptr, /* ctype */

9374

nullptr, /* to_lower */

9375

nullptr, /* to_upper */

9376

nullptr, /* sort_order */

9377

nullptr, /* uca */

9378

nullptr, /* tab_to_uni */

9379

nullptr, /* tab_from_uni */

9380

&my_unicase_default, /* caseinfo */

9381

nullptr, /* state_map */

9382

nullptr, /* ident_map */

9383

8, /* strxfrm_multiply */

9384

1, /* caseup_multiply */

9385

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9390

0xFFFF, /* max_sort_char */

9391

' ', /* pad char */

9392

false, /* escape_with_backslash_is_dangerous */

9393

1, /* levels_for_compare */

9394

&my_charset_utf16_handler,

9395

&my_collation_utf16_uca_handler,

9396

PAD_SPACE};

9397

9398

CHARSET_INFO my_charset_utf16_croatian_uca_ci = {

122,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9403

"utf16", /* cs name */

9404

"utf16_croatian_ci", /* m_coll_name */

9405

"", /* comment */

9406

croatian, /* tailoring */

9407

nullptr, /* coll_param */

9408

nullptr, /* ctype */

9409

nullptr, /* to_lower */

9410

nullptr, /* to_upper */

9411

nullptr, /* sort_order */

9412

nullptr, /* uca */

9413

nullptr, /* tab_to_uni */

9414

nullptr, /* tab_from_uni */

9415

&my_unicase_default, /* caseinfo */

9416

nullptr, /* state_map */

9417

nullptr, /* ident_map */

9418

8, /* strxfrm_multiply */

9419

1, /* caseup_multiply */

9420

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9425

0xFFFF, /* max_sort_char */

9426

' ', /* pad char */

9427

false, /* escape_with_backslash_is_dangerous */

9428

1, /* levels_for_compare */

9429

&my_charset_utf16_handler,

9430

&my_collation_utf16_uca_handler,

9431

PAD_SPACE};

9432

9433

CHARSET_INFO my_charset_utf16_unicode_520_ci = {

123,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_UNICODE | MY_CS_NONASCII,

9438

"utf16", /* csname */

9439

"utf16_unicode_520_ci", /* m_coll_name */

9440

"", /* comment */

9441

"", /* tailoring */

9442

nullptr, /* coll_param */

9443

nullptr, /* ctype */

9444

nullptr, /* to_lower */

9445

nullptr, /* to_upper */

9446

nullptr, /* sort_order */

9447

&my_uca_v520, /* uca */

9448

nullptr, /* tab_to_uni */

9449

nullptr, /* tab_from_uni */

9450

&my_unicase_unicode520, /* caseinfo */

9451

nullptr, /* state_map */

9452

nullptr, /* ident_map */

9453

8, /* strxfrm_multiply */

9454

1, /* caseup_multiply */

9455

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9460

0x10FFFF, /* max_sort_char */

9461

0x20, /* pad char */

9462

false, /* escape_with_backslash_is_dangerous */

9463

1, /* levels_for_compare */

9464

&my_charset_utf16_handler,

9465

&my_collation_utf16_uca_handler,

9466

PAD_SPACE};

9467

9468

CHARSET_INFO my_charset_utf16_vietnamese_ci = {

124,

0,

0, /* number */

MY_CS_UTF16_UCA_FLAGS, /* state */

9473

"utf16", /* cs name */

9474

"utf16_vietnamese_ci", /* m_coll_name */

9475

"", /* comment */

9476

vietnamese, /* tailoring */

9477

nullptr, /* coll_param */

9478

nullptr, /* ctype */

9479

nullptr, /* to_lower */

9480

nullptr, /* to_upper */

9481

nullptr, /* sort_order */

9482

nullptr, /* uca */

9483

nullptr, /* tab_to_uni */

9484

nullptr, /* tab_from_uni */

9485

&my_unicase_default, /* caseinfo */

9486

nullptr, /* state_map */

9487

nullptr, /* ident_map */

9488

8, /* strxfrm_multiply */

9489

1, /* caseup_multiply */

9490

1, /* casedn_multiply */

2, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9495

0xFFFF, /* max_sort_char */

9496

' ', /* pad char */

9497

false, /* escape_with_backslash_is_dangerous */

9498

1, /* levels_for_compare */

9499

&my_charset_utf16_handler,

9500

&my_collation_utf16_uca_handler,

9501

PAD_SPACE};

9502

9503

MY_COLLATION_HANDLER my_collation_gb18030_uca_handler = {

9504

my_coll_init_uca, /* init */

9505

my_coll_uninit_uca,

9506

my_strnncoll_any_uca,

9507

my_strnncollsp_any_uca,

9508

my_strnxfrm_any_uca,

9509

my_strnxfrmlen_simple,

my_like_range_mb,

my_wildcmp_uca,

nullptr,

my_instr_mb,

my_hash_sort_any_uca,

9515

my_propagate_complex};

9516

9517

/**

9518

The array used for "type of characters" bit mask for each

9519

character. The ctype[0] is reserved for EOF(-1), so we use

9520

ctype[(char)+1]. Also refer to strings/CHARSET_INFO.txt

9521

*/

9522

static const uchar ctype_gb18030[257] = {

9523

0, /* For standard library */

9524

32, 32, 32, 32, 32, 32, 32, 32, 32, 40, 40, 40, 40, 40, 32, 32,

9525

32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32,

9526

72, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16,

9527

132, 132, 132, 132, 132, 132, 132, 132, 132, 132, 16, 16, 16, 16, 16, 16,

9528

16, 129, 129, 129, 129, 129, 129, 1, 1, 1, 1, 1, 1, 1, 1, 1,

9529

1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 16, 16, 16, 16, 16,

9530

16, 130, 130, 130, 130, 130, 130, 2, 2, 2, 2, 2, 2, 2, 2, 2,

9531

2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 16, 16, 16, 16, 32,

9532

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9533

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9534

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9535

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9536

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9537

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9538

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,

9539

3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 0};

9540

9541

extern MY_CHARSET_HANDLER my_charset_gb18030_uca_handler;

9542

9543

CHARSET_INFO my_charset_gb18030_unicode_520_ci = {

250,

0,

0, /* number */

MY_CS_COMPILED | MY_CS_STRNXFRM | MY_CS_NONASCII, /* state */

9548

"gb18030", /* cs name */

9549

"gb18030_unicode_520_ci", /* m_coll_name */

9550

"China National Standard GB18030", /* comment */

9551

"", /* tailoring */

9552

nullptr, /* coll_param */

9553

ctype_gb18030, /* ctype */

9554

nullptr, /* lower */

9555

nullptr, /* UPPER */

9556

nullptr, /* sort order */

9557

&my_uca_v520, /* uca */

9558

nullptr, /* tab_to_uni */

9559

nullptr, /* tab_from_uni */

9560

&my_unicase_unicode520, /* caseinfo */

9561

nullptr, /* state_map */

9562

nullptr, /* ident_map */

9563

8, /* strxfrm_multiply */

9564

2, /* caseup_multiply */

9565

2, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

2, /* mbmaxlenlen */

0, /* min_sort_char */

9570

0xE3329A35, /* max_sort_char */

9571

' ', /* pad char */

9572

false, /* escape_with_backslash_is_dangerous */

9573

1, /* levels_for_compare */

9574

&my_charset_gb18030_uca_handler,

9575

&my_collation_gb18030_uca_handler,

9576

PAD_SPACE};

9577

9578

CHARSET_INFO my_charset_utf8mb4_0900_ai_ci = {

255,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_PRIMARY, /* state */

9583

MY_UTF8MB4, /* csname */

9584

MY_UTF8MB4 "_0900_ai_ci", /* m_coll_name */

9585

"UTF-8 Unicode", /* comment */

9586

nullptr, /* tailoring */

9587

nullptr, /* coll_param */

9588

ctype_utf8, /* ctype */

9589

nullptr, /* to_lower */

9590

nullptr, /* to_upper */

9591

nullptr, /* sort_order */

9592

&my_uca_v900, /* uca_900 */

9593

nullptr, /* tab_to_uni */

9594

nullptr, /* tab_from_uni */

9595

&my_unicase_unicode900, /* caseinfo */

9596

nullptr, /* state_map */

9597

nullptr, /* ident_map */

9598

0, /* strxfrm_multiply */

9599

1, /* caseup_multiply */

9600

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9605

0x10FFFF, /* max_sort_char */

9606

' ', /* pad char */

9607

false, /* escape_with_backslash_is_dangerous */

9608

1, /* levels_for_compare */

9609

&my_charset_utf8mb4_handler,

9610

&my_collation_uca_900_handler,

9611

NO_PAD};

9612

9613

CHARSET_INFO my_charset_utf8mb4_de_pb_0900_ai_ci = {

256,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9618

MY_UTF8MB4, /* csname */

9619

MY_UTF8MB4 "_de_pb_0900_ai_ci", /* m_coll_name */

9620

"", /* comment */

9621

de_pb_cldr_30, /* tailoring */

9622

nullptr, /* coll_param */

9623

ctype_utf8, /* ctype */

9624

nullptr, /* to_lower */

9625

nullptr, /* to_upper */

9626

nullptr, /* sort_order */

9627

&my_uca_v900, /* uca_900 */

9628

nullptr, /* tab_to_uni */

9629

nullptr, /* tab_from_uni */

9630

&my_unicase_unicode900, /* caseinfo */

9631

nullptr, /* state_map */

9632

nullptr, /* ident_map */

9633

0, /* strxfrm_multiply */

9634

1, /* caseup_multiply */

9635

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9640

0x10FFFF, /* max_sort_char */

9641

' ', /* pad char */

9642

false, /* escape_with_backslash_is_dangerous */

9643

1, /* levels_for_compare */

9644

&my_charset_utf8mb4_handler,

9645

&my_collation_uca_900_handler,

9646

NO_PAD};

9647

9648

CHARSET_INFO my_charset_utf8mb4_is_0900_ai_ci = {

257,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9653

MY_UTF8MB4, /* csname */

9654

MY_UTF8MB4 "_is_0900_ai_ci", /* m_coll_name */

9655

"", /* comment */

9656

is_cldr_30, /* tailoring */

9657

nullptr, /* coll_param */

9658

ctype_utf8, /* ctype */

9659

nullptr, /* to_lower */

9660

nullptr, /* to_upper */

9661

nullptr, /* sort_order */

9662

&my_uca_v900, /* uca */

9663

nullptr, /* tab_to_uni */

9664

nullptr, /* tab_from_uni */

9665

&my_unicase_unicode900, /* caseinfo */

9666

nullptr, /* state_map */

9667

nullptr, /* ident_map */

9668

0, /* strxfrm_multiply */

9669

1, /* caseup_multiply */

9670

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9675

0x10FFFF, /* max_sort_char */

9676

' ', /* pad char */

9677

false, /* escape_with_backslash_is_dangerous */

9678

1, /* levels_for_compare */

9679

&my_charset_utf8mb4_handler,

9680

&my_collation_uca_900_handler,

9681

NO_PAD};

9682

9683

CHARSET_INFO my_charset_utf8mb4_lv_0900_ai_ci = {

258,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9688

MY_UTF8MB4, /* csname */

9689

MY_UTF8MB4 "_lv_0900_ai_ci", /* m_coll_name */

9690

"", /* comment */

9691

lv_cldr_30, /* tailoring */

9692

nullptr, /* coll_param */

9693

ctype_utf8, /* ctype */

9694

nullptr, /* to_lower */

9695

nullptr, /* to_upper */

9696

nullptr, /* sort_order */

9697

&my_uca_v900, /* uca */

9698

nullptr, /* tab_to_uni */

9699

nullptr, /* tab_from_uni */

9700

&my_unicase_unicode900, /* caseinfo */

9701

nullptr, /* state_map */

9702

nullptr, /* ident_map */

9703

0, /* strxfrm_multiply */

9704

1, /* caseup_multiply */

9705

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9710

0x10FFFF, /* max_sort_char */

9711

' ', /* pad char */

9712

false, /* escape_with_backslash_is_dangerous */

9713

1, /* levels_for_compare */

9714

&my_charset_utf8mb4_handler,

9715

&my_collation_uca_900_handler,

9716

NO_PAD};

9717

9718

CHARSET_INFO my_charset_utf8mb4_ro_0900_ai_ci = {

259,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9723

MY_UTF8MB4, /* csname */

9724

MY_UTF8MB4 "_ro_0900_ai_ci", /* m_coll_name */

9725

"", /* comment */

9726

ro_cldr_30, /* tailoring */

9727

nullptr, /* coll_param */

9728

ctype_utf8, /* ctype */

9729

nullptr, /* to_lower */

9730

nullptr, /* to_upper */

9731

nullptr, /* sort_order */

9732

&my_uca_v900, /* uca */

9733

nullptr, /* tab_to_uni */

9734

nullptr, /* tab_from_uni */

9735

&my_unicase_unicode900, /* caseinfo */

9736

nullptr, /* state_map */

9737

nullptr, /* ident_map */

9738

0, /* strxfrm_multiply */

9739

1, /* caseup_multiply */

9740

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9745

0x10FFFF, /* max_sort_char */

9746

' ', /* pad char */

9747

false, /* escape_with_backslash_is_dangerous */

9748

1, /* levels_for_compare */

9749

&my_charset_utf8mb4_handler,

9750

&my_collation_uca_900_handler,

9751

NO_PAD};

9752

9753

CHARSET_INFO my_charset_utf8mb4_sl_0900_ai_ci = {

260,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9758

MY_UTF8MB4, /* csname */

9759

MY_UTF8MB4 "_sl_0900_ai_ci", /* m_coll_name */

9760

"", /* comment */

9761

sl_cldr_30, /* tailoring */

9762

nullptr, /* coll_param */

9763

ctype_utf8, /* ctype */

9764

nullptr, /* to_lower */

9765

nullptr, /* to_upper */

9766

nullptr, /* sort_order */

9767

&my_uca_v900, /* uca */

9768

nullptr, /* tab_to_uni */

9769

nullptr, /* tab_from_uni */

9770

&my_unicase_unicode900, /* caseinfo */

9771

nullptr, /* state_map */

9772

nullptr, /* ident_map */

9773

0, /* strxfrm_multiply */

9774

1, /* caseup_multiply */

9775

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9780

0x10FFFF, /* max_sort_char */

9781

' ', /* pad char */

9782

false, /* escape_with_backslash_is_dangerous */

9783

1, /* levels_for_compare */

9784

&my_charset_utf8mb4_handler,

9785

&my_collation_uca_900_handler,

9786

NO_PAD};

9787

9788

CHARSET_INFO my_charset_utf8mb4_pl_0900_ai_ci = {

261,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9793

MY_UTF8MB4, /* csname */

9794

MY_UTF8MB4 "_pl_0900_ai_ci", /* m_coll_name */

9795

"", /* comment */

9796

pl_cldr_30, /* tailoring */

9797

nullptr, /* coll_param */

9798

ctype_utf8, /* ctype */

9799

nullptr, /* to_lower */

9800

nullptr, /* to_upper */

9801

nullptr, /* sort_order */

9802

&my_uca_v900, /* uca */

9803

nullptr, /* tab_to_uni */

9804

nullptr, /* tab_from_uni */

9805

&my_unicase_unicode900, /* caseinfo */

9806

nullptr, /* state_map */

9807

nullptr, /* ident_map */

9808

0, /* strxfrm_multiply */

9809

1, /* caseup_multiply */

9810

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9815

0x10FFFF, /* max_sort_char */

9816

' ', /* pad char */

9817

false, /* escape_with_backslash_is_dangerous */

9818

1, /* levels_for_compare */

9819

&my_charset_utf8mb4_handler,

9820

&my_collation_uca_900_handler,

9821

NO_PAD};

9822

9823

CHARSET_INFO my_charset_utf8mb4_et_0900_ai_ci = {

262,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9828

MY_UTF8MB4, /* csname */

9829

MY_UTF8MB4 "_et_0900_ai_ci", /* m_coll_name */

9830

"", /* comment */

9831

et_cldr_30, /* tailoring */

9832

nullptr, /* coll_param */

9833

ctype_utf8, /* ctype */

9834

nullptr, /* to_lower */

9835

nullptr, /* to_upper */

9836

nullptr, /* sort_order */

9837

&my_uca_v900, /* uca */

9838

nullptr, /* tab_to_uni */

9839

nullptr, /* tab_from_uni */

9840

&my_unicase_unicode900, /* caseinfo */

9841

nullptr, /* state_map */

9842

nullptr, /* ident_map */

9843

0, /* strxfrm_multiply */

9844

1, /* caseup_multiply */

9845

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9850

0x10FFFF, /* max_sort_char */

9851

' ', /* pad char */

9852

false, /* escape_with_backslash_is_dangerous */

9853

1, /* levels_for_compare */

9854

&my_charset_utf8mb4_handler,

9855

&my_collation_uca_900_handler,

9856

NO_PAD};

9857

9858

CHARSET_INFO my_charset_utf8mb4_es_0900_ai_ci = {

263,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9863

MY_UTF8MB4, /* csname */

9864

MY_UTF8MB4 "_es_0900_ai_ci", /* m_coll_name */

9865

"", /* comment */

9866

spanish, /* tailoring */

9867

nullptr, /* coll_param */

9868

ctype_utf8, /* ctype */

9869

nullptr, /* to_lower */

9870

nullptr, /* to_upper */

9871

nullptr, /* sort_order */

9872

&my_uca_v900, /* uca */

9873

nullptr, /* tab_to_uni */

9874

nullptr, /* tab_from_uni */

9875

&my_unicase_unicode900, /* caseinfo */

9876

nullptr, /* state_map */

9877

nullptr, /* ident_map */

9878

0, /* strxfrm_multiply */

9879

1, /* caseup_multiply */

9880

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9885

0x10FFFF, /* max_sort_char */

9886

' ', /* pad char */

9887

false, /* escape_with_backslash_is_dangerous */

9888

1, /* levels_for_compare */

9889

&my_charset_utf8mb4_handler,

9890

&my_collation_uca_900_handler,

9891

NO_PAD};

9892

9893

CHARSET_INFO my_charset_utf8mb4_sv_0900_ai_ci = {

264,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9898

MY_UTF8MB4, /* csname */

9899

MY_UTF8MB4 "_sv_0900_ai_ci", /* m_coll_name */

9900

"", /* comment */

9901

sv_cldr_30, /* tailoring */

9902

nullptr, /* coll_param */

9903

ctype_utf8, /* ctype */

9904

nullptr, /* to_lower */

9905

nullptr, /* to_upper */

9906

nullptr, /* sort_order */

9907

&my_uca_v900, /* uca */

9908

nullptr, /* tab_to_uni */

9909

nullptr, /* tab_from_uni */

9910

&my_unicase_unicode900, /* caseinfo */

9911

nullptr, /* state_map */

9912

nullptr, /* ident_map */

9913

0, /* strxfrm_multiply */

9914

1, /* caseup_multiply */

9915

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9920

0x10FFFF, /* max_sort_char */

9921

' ', /* pad char */

9922

false, /* escape_with_backslash_is_dangerous */

9923

1, /* levels_for_compare */

9924

&my_charset_utf8mb4_handler,

9925

&my_collation_uca_900_handler,

9926

NO_PAD};

9927

9928

CHARSET_INFO my_charset_utf8mb4_tr_0900_ai_ci = {

265,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9933

MY_UTF8MB4, /* csname */

9934

MY_UTF8MB4 "_tr_0900_ai_ci", /* m_coll_name */

9935

"", /* comment */

9936

tr_cldr_30, /* tailoring */

9937

nullptr, /* coll_param */

9938

ctype_utf8, /* ctype */

9939

nullptr, /* to_lower */

9940

nullptr, /* to_upper */

9941

nullptr, /* sort_order */

9942

&my_uca_v900, /* uca */

9943

nullptr, /* tab_to_uni */

9944

nullptr, /* tab_from_uni */

9945

&my_unicase_unicode900, /* caseinfo */

9946

nullptr, /* state_map */

9947

nullptr, /* ident_map */

9948

0, /* strxfrm_multiply */

9949

1, /* caseup_multiply */

9950

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9955

0x10FFFF, /* max_sort_char */

9956

' ', /* pad char */

9957

false, /* escape_with_backslash_is_dangerous */

9958

1, /* levels_for_compare */

9959

&my_charset_utf8mb4_handler,

9960

&my_collation_uca_900_handler,

9961

NO_PAD};

9962

9963

CHARSET_INFO my_charset_utf8mb4_cs_0900_ai_ci = {

266,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

9968

MY_UTF8MB4, /* csname */

9969

MY_UTF8MB4 "_cs_0900_ai_ci", /* m_coll_name */

9970

"", /* comment */

9971

cs_cldr_30, /* tailoring */

9972

nullptr, /* coll_param */

9973

ctype_utf8, /* ctype */

9974

nullptr, /* to_lower */

9975

nullptr, /* to_upper */

9976

nullptr, /* sort_order */

9977

&my_uca_v900, /* uca */

9978

nullptr, /* tab_to_uni */

9979

nullptr, /* tab_from_uni */

9980

&my_unicase_unicode900, /* caseinfo */

9981

nullptr, /* state_map */

9982

nullptr, /* ident_map */

9983

0, /* strxfrm_multiply */

9984

1, /* caseup_multiply */

9985

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

9990

0x10FFFF, /* max_sort_char */

9991

' ', /* pad char */

9992

false, /* escape_with_backslash_is_dangerous */

9993

1, /* levels_for_compare */

9994

&my_charset_utf8mb4_handler,

9995

&my_collation_uca_900_handler,

9996

NO_PAD};

9997

9998

CHARSET_INFO my_charset_utf8mb4_da_0900_ai_ci = {

267,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10003

MY_UTF8MB4, /* csname */

10004

MY_UTF8MB4 "_da_0900_ai_ci", /* m_coll_name */

10005

"", /* comment */

10006

da_cldr_30, /* tailoring */

10007

nullptr, /* coll_param */

10008

ctype_utf8, /* ctype */

10009

nullptr, /* to_lower */

10010

nullptr, /* to_upper */

10011

nullptr, /* sort_order */

10012

&my_uca_v900, /* uca */

10013

nullptr, /* tab_to_uni */

10014

nullptr, /* tab_from_uni */

10015

&my_unicase_unicode900, /* caseinfo */

10016

nullptr, /* state_map */

10017

nullptr, /* ident_map */

10018

0, /* strxfrm_multiply */

10019

1, /* caseup_multiply */

10020

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10025

0x10FFFF, /* max_sort_char */

10026

' ', /* pad char */

10027

false, /* escape_with_backslash_is_dangerous */

10028

1, /* levels_for_compare */

10029

&my_charset_utf8mb4_handler,

10030

&my_collation_uca_900_handler,

10031

NO_PAD};

10032

10033

CHARSET_INFO my_charset_utf8mb4_lt_0900_ai_ci = {

268,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10038

MY_UTF8MB4, /* csname */

10039

MY_UTF8MB4 "_lt_0900_ai_ci", /* m_coll_name */

10040

"", /* comment */

10041

lt_cldr_30, /* tailoring */

10042

nullptr, /* coll_param */

10043

ctype_utf8, /* ctype */

10044

nullptr, /* to_lower */

10045

nullptr, /* to_upper */

10046

nullptr, /* sort_order */

10047

&my_uca_v900, /* uca */

10048

nullptr, /* tab_to_uni */

10049

nullptr, /* tab_from_uni */

10050

&my_unicase_unicode900, /* caseinfo */

10051

nullptr, /* state_map */

10052

nullptr, /* ident_map */

10053

0, /* strxfrm_multiply */

10054

1, /* caseup_multiply */

10055

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10060

0x10FFFF, /* max_sort_char */

10061

' ', /* pad char */

10062

false, /* escape_with_backslash_is_dangerous */

10063

1, /* levels_for_compare */

10064

&my_charset_utf8mb4_handler,

10065

&my_collation_uca_900_handler,

10066

NO_PAD};

10067

10068

CHARSET_INFO my_charset_utf8mb4_sk_0900_ai_ci = {

269,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10073

MY_UTF8MB4, /* csname */

10074

MY_UTF8MB4 "_sk_0900_ai_ci", /* m_coll_name */

10075

"", /* comment */

10076

sk_cldr_30, /* tailoring */

10077

nullptr, /* coll_param */

10078

ctype_utf8, /* ctype */

10079

nullptr, /* to_lower */

10080

nullptr, /* to_upper */

10081

nullptr, /* sort_order */

10082

&my_uca_v900, /* uca */

10083

nullptr, /* tab_to_uni */

10084

nullptr, /* tab_from_uni */

10085

&my_unicase_unicode900, /* caseinfo */

10086

nullptr, /* state_map */

10087

nullptr, /* ident_map */

10088

0, /* strxfrm_multiply */

10089

1, /* caseup_multiply */

10090

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10095

0x10FFFF, /* max_sort_char */

10096

' ', /* pad char */

10097

false, /* escape_with_backslash_is_dangerous */

10098

1, /* levels_for_compare */

10099

&my_charset_utf8mb4_handler,

10100

&my_collation_uca_900_handler,

10101

NO_PAD};

10102

10103

CHARSET_INFO my_charset_utf8mb4_es_trad_0900_ai_ci = {

270,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10108

MY_UTF8MB4, /* csname */

10109

MY_UTF8MB4 "_es_trad_0900_ai_ci", /* m_coll_name */

10110

"", /* comment */

10111

es_trad_cldr_30, /* tailoring */

10112

nullptr, /* coll_param */

10113

ctype_utf8, /* ctype */

10114

nullptr, /* to_lower */

10115

nullptr, /* to_upper */

10116

nullptr, /* sort_order */

10117

&my_uca_v900, /* uca */

10118

nullptr, /* tab_to_uni */

10119

nullptr, /* tab_from_uni */

10120

&my_unicase_unicode900, /* caseinfo */

10121

nullptr, /* state_map */

10122

nullptr, /* ident_map */

10123

0, /* strxfrm_multiply */

10124

1, /* caseup_multiply */

10125

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10130

0x10FFFF, /* max_sort_char */

10131

' ', /* pad char */

10132

false, /* escape_with_backslash_is_dangerous */

10133

1, /* levels_for_compare */

10134

&my_charset_utf8mb4_handler,

10135

&my_collation_uca_900_handler,

10136

NO_PAD};

10137

10138

CHARSET_INFO my_charset_utf8mb4_la_0900_ai_ci = {

271,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10143

MY_UTF8MB4, /* csname */

10144

MY_UTF8MB4 "_la_0900_ai_ci", /* m_coll_name */

10145

"", /* comment */

10146

roman, /* tailoring */

10147

nullptr, /* coll_param */

10148

ctype_utf8, /* ctype */

10149

nullptr, /* to_lower */

10150

nullptr, /* to_upper */

10151

nullptr, /* sort_order */

10152

&my_uca_v900, /* uca */

10153

nullptr, /* tab_to_uni */

10154

nullptr, /* tab_from_uni */

10155

&my_unicase_unicode900, /* caseinfo */

10156

nullptr, /* state_map */

10157

nullptr, /* ident_map */

10158

0, /* strxfrm_multiply */

10159

1, /* caseup_multiply */

10160

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10165

0x10FFFF, /* max_sort_char */

10166

' ', /* pad char */

10167

false, /* escape_with_backslash_is_dangerous */

10168

1, /* levels_for_compare */

10169

&my_charset_utf8mb4_handler,

10170

&my_collation_uca_900_handler,

NO_PAD};

#if 0

CHARSET_INFO my_charset_utf8mb4_fa_0900_ai_ci=

10175

{

10176

272, 0, 0, /* number */

10177

MY_CS_UTF8MB4_UCA_FLAGS,/* state */

10178

MY_UTF8MB4, /* csname */

10179

MY_UTF8MB4 "_fa_0900_ai_ci",/* m_coll_name */

10180

"", /* comment */

10181

fa_cldr_30, /* tailoring */

10182

&fa_coll_param, /* coll_param */

10183

ctype_utf8, /* ctype */

10184

NULL, /* to_lower */

10185

NULL, /* to_upper */

10186

NULL, /* sort_order */

10187

&my_uca_v900, /* uca */

10188

NULL, /* tab_to_uni */

10189

NULL, /* tab_from_uni */

10190

&my_unicase_unicode900,/* caseinfo */

10191

NULL, /* state_map */

10192

NULL, /* ident_map */

10193

0, /* strxfrm_multiply */

10194

1, /* caseup_multiply */

10195

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10200

0x10FFFF, /* max_sort_char */

10201

' ', /* pad char */

10202

0, /* escape_with_backslash_is_dangerous */

10203

1, /* levels_for_compare */

10204

&my_charset_utf8mb4_handler,

10205

&my_collation_uca_900_handler,

NO_PAD

};

#endif

CHARSET_INFO my_charset_utf8mb4_eo_0900_ai_ci = {

273,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10215

MY_UTF8MB4, /* csname */

10216

MY_UTF8MB4 "_eo_0900_ai_ci", /* m_coll_name */

10217

"", /* comment */

10218

esperanto, /* tailoring */

10219

nullptr, /* coll_param */

10220

ctype_utf8, /* ctype */

10221

nullptr, /* to_lower */

10222

nullptr, /* to_upper */

10223

nullptr, /* sort_order */

10224

&my_uca_v900, /* uca */

10225

nullptr, /* tab_to_uni */

10226

nullptr, /* tab_from_uni */

10227

&my_unicase_unicode900, /* caseinfo */

10228

nullptr, /* state_map */

10229

nullptr, /* ident_map */

10230

0, /* strxfrm_multiply */

10231

1, /* caseup_multiply */

10232

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10237

0x10FFFF, /* max_sort_char */

10238

' ', /* pad char */

10239

false, /* escape_with_backslash_is_dangerous */

10240

1, /* levels_for_compare */

10241

&my_charset_utf8mb4_handler,

10242

&my_collation_uca_900_handler,

10243

NO_PAD};

10244

10245

CHARSET_INFO my_charset_utf8mb4_hu_0900_ai_ci = {

274,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10250

MY_UTF8MB4, /* csname */

10251

MY_UTF8MB4 "_hu_0900_ai_ci", /* m_coll_name */

10252

"", /* comment */

10253

hu_cldr_30, /* tailoring */

10254

nullptr, /* coll_param */

10255

ctype_utf8, /* ctype */

10256

nullptr, /* to_lower */

10257

nullptr, /* to_upper */

10258

nullptr, /* sort_order */

10259

&my_uca_v900, /* uca */

10260

nullptr, /* tab_to_uni */

10261

nullptr, /* tab_from_uni */

10262

&my_unicase_unicode900, /* caseinfo */

10263

nullptr, /* state_map */

10264

nullptr, /* ident_map */

10265

0, /* strxfrm_multiply */

10266

1, /* caseup_multiply */

10267

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10272

0x10FFFF, /* max_sort_char */

10273

' ', /* pad char */

10274

false, /* escape_with_backslash_is_dangerous */

10275

1, /* levels_for_compare */

10276

&my_charset_utf8mb4_handler,

10277

&my_collation_uca_900_handler,

10278

NO_PAD};

10279

10280

CHARSET_INFO my_charset_utf8mb4_hr_0900_ai_ci = {

275,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10285

MY_UTF8MB4, /* csname */

10286

MY_UTF8MB4 "_hr_0900_ai_ci", /* m_coll_name */

10287

"", /* comment */

10288

hr_cldr_30, /* tailoring */

10289

&hr_coll_param, /* coll_param */

10290

ctype_utf8, /* ctype */

10291

nullptr, /* to_lower */

10292

nullptr, /* to_upper */

10293

nullptr, /* sort_order */

10294

&my_uca_v900, /* uca */

10295

nullptr, /* tab_to_uni */

10296

nullptr, /* tab_from_uni */

10297

&my_unicase_unicode900, /* caseinfo */

10298

nullptr, /* state_map */

10299

nullptr, /* ident_map */

10300

0, /* strxfrm_multiply */

10301

1, /* caseup_multiply */

10302

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10307

0x10FFFF, /* max_sort_char */

10308

' ', /* pad char */

10309

false, /* escape_with_backslash_is_dangerous */

10310

1, /* levels_for_compare */

10311

&my_charset_utf8mb4_handler,

10312

&my_collation_uca_900_handler,

NO_PAD};

#if 0

CHARSET_INFO my_charset_utf8mb4_si_0900_ai_ci=

10317

{

10318

276, 0, 0, /* number */

10319

MY_CS_UTF8MB4_UCA_FLAGS,/* state */

10320

MY_UTF8MB4, /* csname */

10321

MY_UTF8MB4 "_si_0900_ai_ci",/* m_coll_name */

10322

"", /* comment */

10323

si_cldr_30, /* tailoring */

10324

NULL, /* coll_param */

10325

ctype_utf8, /* ctype */

10326

NULL, /* to_lower */

10327

NULL, /* to_upper */

10328

NULL, /* sort_order */

10329

&my_uca_v900, /* uca */

10330

NULL, /* tab_to_uni */

10331

NULL, /* tab_from_uni */

10332

&my_unicase_unicode900,/* caseinfo */

10333

NULL, /* state_map */

10334

NULL, /* ident_map */

10335

0, /* strxfrm_multiply */

10336

1, /* caseup_multiply */

10337

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10342

0x10FFFF, /* max_sort_char */

10343

' ', /* pad char */

10344

0, /* escape_with_backslash_is_dangerous */

10345

1, /* levels_for_compare */

10346

&my_charset_utf8mb4_handler,

10347

&my_collation_uca_900_handler,

NO_PAD

};

#endif

CHARSET_INFO my_charset_utf8mb4_vi_0900_ai_ci = {

277,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

10357

MY_UTF8MB4, /* csname */

10358

MY_UTF8MB4 "_vi_0900_ai_ci", /* m_coll_name */

10359

"", /* comment */

10360

vi_cldr_30, /* tailoring */

10361

nullptr, /* coll_param */

10362

ctype_utf8, /* ctype */

10363

nullptr, /* to_lower */

10364

nullptr, /* to_upper */

10365

nullptr, /* sort_order */

10366

&my_uca_v900, /* uca */

10367

nullptr, /* tab_to_uni */

10368

nullptr, /* tab_from_uni */

10369

&my_unicase_unicode900, /* caseinfo */

10370

nullptr, /* state_map */

10371

nullptr, /* ident_map */

10372

0, /* strxfrm_multiply */

10373

1, /* caseup_multiply */

10374

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10379

0x10FFFF, /* max_sort_char */

10380

' ', /* pad char */

10381

false, /* escape_with_backslash_is_dangerous */

10382

1, /* levels_for_compare */

10383

&my_charset_utf8mb4_handler,

10384

&my_collation_uca_900_handler,

10385

NO_PAD};

10386

10387

CHARSET_INFO my_charset_utf8mb4_0900_as_cs = {

278,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10392

MY_UTF8MB4, /* csname */

10393

MY_UTF8MB4 "_0900_as_cs", /* m_coll_name */

10394

"", /* comment */

10395

nullptr, /* tailoring */

10396

nullptr, /* coll_param */

10397

ctype_utf8, /* ctype */

10398

nullptr, /* to_lower */

10399

nullptr, /* to_upper */

10400

nullptr, /* sort_order */

10401

&my_uca_v900, /* uca */

10402

nullptr, /* tab_to_uni */

10403

nullptr, /* tab_from_uni */

10404

&my_unicase_unicode900, /* caseinfo */

10405

nullptr, /* state_map */

10406

nullptr, /* ident_map */

10407

0, /* strxfrm_multiply */

10408

1, /* caseup_multiply */

10409

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10414

0x10FFFF, /* max_sort_char */

10415

' ', /* pad char */

10416

false, /* escape_with_backslash_is_dangerous */

10417

3, /* levels_for_compare */

10418

&my_charset_utf8mb4_handler,

10419

&my_collation_uca_900_handler,

10420

NO_PAD};

10421

10422

CHARSET_INFO my_charset_utf8mb4_de_pb_0900_as_cs = {

279,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10427

MY_UTF8MB4, /* csname */

10428

MY_UTF8MB4 "_de_pb_0900_as_cs", /* m_coll_name */

10429

"", /* comment */

10430

de_pb_cldr_30, /* tailoring */

10431

nullptr, /* coll_param */

10432

ctype_utf8, /* ctype */

10433

nullptr, /* to_lower */

10434

nullptr, /* to_upper */

10435

nullptr, /* sort_order */

10436

&my_uca_v900, /* uca */

10437

nullptr, /* tab_to_uni */

10438

nullptr, /* tab_from_uni */

10439

&my_unicase_unicode900, /* caseinfo */

10440

nullptr, /* state_map */

10441

nullptr, /* ident_map */

10442

0, /* strxfrm_multiply */

10443

1, /* caseup_multiply */

10444

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10449

0x10FFFF, /* max_sort_char */

10450

' ', /* pad char */

10451

false, /* escape_with_backslash_is_dangerous */

10452

3, /* levels_for_compare */

10453

&my_charset_utf8mb4_handler,

10454

&my_collation_uca_900_handler,

10455

NO_PAD};

10456

10457

CHARSET_INFO my_charset_utf8mb4_is_0900_as_cs = {

280,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10462

MY_UTF8MB4, /* csname */

10463

MY_UTF8MB4 "_is_0900_as_cs", /* m_coll_name */

10464

"", /* comment */

10465

is_cldr_30, /* tailoring */

10466

nullptr, /* coll_param */

10467

ctype_utf8, /* ctype */

10468

nullptr, /* to_lower */

10469

nullptr, /* to_upper */

10470

nullptr, /* sort_order */

10471

&my_uca_v900, /* uca */

10472

nullptr, /* tab_to_uni */

10473

nullptr, /* tab_from_uni */

10474

&my_unicase_unicode900, /* caseinfo */

10475

nullptr, /* state_map */

10476

nullptr, /* ident_map */

10477

0, /* strxfrm_multiply */

10478

1, /* caseup_multiply */

10479

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10484

0x10FFFF, /* max_sort_char */

10485

' ', /* pad char */

10486

false, /* escape_with_backslash_is_dangerous */

10487

3, /* levels_for_compare */

10488

&my_charset_utf8mb4_handler,

10489

&my_collation_uca_900_handler,

10490

NO_PAD};

10491

10492

CHARSET_INFO my_charset_utf8mb4_lv_0900_as_cs = {

281,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10497

MY_UTF8MB4, /* csname */

10498

MY_UTF8MB4 "_lv_0900_as_cs", /* m_coll_name */

10499

"", /* comment */

10500

lv_cldr_30, /* tailoring */

10501

nullptr, /* coll_param */

10502

ctype_utf8, /* ctype */

10503

nullptr, /* to_lower */

10504

nullptr, /* to_upper */

10505

nullptr, /* sort_order */

10506

&my_uca_v900, /* uca */

10507

nullptr, /* tab_to_uni */

10508

nullptr, /* tab_from_uni */

10509

&my_unicase_unicode900, /* caseinfo */

10510

nullptr, /* state_map */

10511

nullptr, /* ident_map */

10512

0, /* strxfrm_multiply */

10513

1, /* caseup_multiply */

10514

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10519

0x10FFFF, /* max_sort_char */

10520

' ', /* pad char */

10521

false, /* escape_with_backslash_is_dangerous */

10522

3, /* levels_for_compare */

10523

&my_charset_utf8mb4_handler,

10524

&my_collation_uca_900_handler,

10525

NO_PAD};

10526

10527

CHARSET_INFO my_charset_utf8mb4_ro_0900_as_cs = {

282,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10532

MY_UTF8MB4, /* csname */

10533

MY_UTF8MB4 "_ro_0900_as_cs", /* m_coll_name */

10534

"", /* comment */

10535

ro_cldr_30, /* tailoring */

10536

nullptr, /* coll_param */

10537

ctype_utf8, /* ctype */

10538

nullptr, /* to_lower */

10539

nullptr, /* to_upper */

10540

nullptr, /* sort_order */

10541

&my_uca_v900, /* uca */

10542

nullptr, /* tab_to_uni */

10543

nullptr, /* tab_from_uni */

10544

&my_unicase_unicode900, /* caseinfo */

10545

nullptr, /* state_map */

10546

nullptr, /* ident_map */

10547

0, /* strxfrm_multiply */

10548

1, /* caseup_multiply */

10549

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10554

0x10FFFF, /* max_sort_char */

10555

' ', /* pad char */

10556

false, /* escape_with_backslash_is_dangerous */

10557

3, /* levels_for_compare */

10558

&my_charset_utf8mb4_handler,

10559

&my_collation_uca_900_handler,

10560

NO_PAD};

10561

10562

CHARSET_INFO my_charset_utf8mb4_sl_0900_as_cs = {

283,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10567

MY_UTF8MB4, /* csname */

10568

MY_UTF8MB4 "_sl_0900_as_cs", /* m_coll_name */

10569

"", /* comment */

10570

sl_cldr_30, /* tailoring */

10571

nullptr, /* coll_param */

10572

ctype_utf8, /* ctype */

10573

nullptr, /* to_lower */

10574

nullptr, /* to_upper */

10575

nullptr, /* sort_order */

10576

&my_uca_v900, /* uca */

10577

nullptr, /* tab_to_uni */

10578

nullptr, /* tab_from_uni */

10579

&my_unicase_unicode900, /* caseinfo */

10580

nullptr, /* state_map */

10581

nullptr, /* ident_map */

10582

0, /* strxfrm_multiply */

10583

1, /* caseup_multiply */

10584

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10589

0x10FFFF, /* max_sort_char */

10590

' ', /* pad char */

10591

false, /* escape_with_backslash_is_dangerous */

10592

3, /* levels_for_compare */

10593

&my_charset_utf8mb4_handler,

10594

&my_collation_uca_900_handler,

10595

NO_PAD};

10596

10597

CHARSET_INFO my_charset_utf8mb4_pl_0900_as_cs = {

284,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10602

MY_UTF8MB4, /* csname */

10603

MY_UTF8MB4 "_pl_0900_as_cs", /* m_coll_name */

10604

"", /* comment */

10605

pl_cldr_30, /* tailoring */

10606

nullptr, /* coll_param */

10607

ctype_utf8, /* ctype */

10608

nullptr, /* to_lower */

10609

nullptr, /* to_upper */

10610

nullptr, /* sort_order */

10611

&my_uca_v900, /* uca */

10612

nullptr, /* tab_to_uni */

10613

nullptr, /* tab_from_uni */

10614

&my_unicase_unicode900, /* caseinfo */

10615

nullptr, /* state_map */

10616

nullptr, /* ident_map */

10617

0, /* strxfrm_multiply */

10618

1, /* caseup_multiply */

10619

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10624

0x10FFFF, /* max_sort_char */

10625

' ', /* pad char */

10626

false, /* escape_with_backslash_is_dangerous */

10627

3, /* levels_for_compare */

10628

&my_charset_utf8mb4_handler,

10629

&my_collation_uca_900_handler,

10630

NO_PAD};

10631

10632

CHARSET_INFO my_charset_utf8mb4_et_0900_as_cs = {

285,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10637

MY_UTF8MB4, /* csname */

10638

MY_UTF8MB4 "_et_0900_as_cs", /* m_coll_name */

10639

"", /* comment */

10640

et_cldr_30, /* tailoring */

10641

nullptr, /* coll_param */

10642

ctype_utf8, /* ctype */

10643

nullptr, /* to_lower */

10644

nullptr, /* to_upper */

10645

nullptr, /* sort_order */

10646

&my_uca_v900, /* uca */

10647

nullptr, /* tab_to_uni */

10648

nullptr, /* tab_from_uni */

10649

&my_unicase_unicode900, /* caseinfo */

10650

nullptr, /* state_map */

10651

nullptr, /* ident_map */

10652

0, /* strxfrm_multiply */

10653

1, /* caseup_multiply */

10654

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10659

0x10FFFF, /* max_sort_char */

10660

' ', /* pad char */

10661

false, /* escape_with_backslash_is_dangerous */

10662

3, /* levels_for_compare */

10663

&my_charset_utf8mb4_handler,

10664

&my_collation_uca_900_handler,

10665

NO_PAD};

10666

10667

CHARSET_INFO my_charset_utf8mb4_es_0900_as_cs = {

286,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10672

MY_UTF8MB4, /* csname */

10673

MY_UTF8MB4 "_es_0900_as_cs", /* m_coll_name */

10674

"", /* comment */

10675

spanish, /* tailoring */

10676

nullptr, /* coll_param */

10677

ctype_utf8, /* ctype */

10678

nullptr, /* to_lower */

10679

nullptr, /* to_upper */

10680

nullptr, /* sort_order */

10681

&my_uca_v900, /* uca */

10682

nullptr, /* tab_to_uni */

10683

nullptr, /* tab_from_uni */

10684

&my_unicase_unicode900, /* caseinfo */

10685

nullptr, /* state_map */

10686

nullptr, /* ident_map */

10687

0, /* strxfrm_multiply */

10688

1, /* caseup_multiply */

10689

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10694

0x10FFFF, /* max_sort_char */

10695

' ', /* pad char */

10696

false, /* escape_with_backslash_is_dangerous */

10697

3, /* levels_for_compare */

10698

&my_charset_utf8mb4_handler,

10699

&my_collation_uca_900_handler,

10700

NO_PAD};

10701

10702

CHARSET_INFO my_charset_utf8mb4_sv_0900_as_cs = {

287,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10707

MY_UTF8MB4, /* csname */

10708

MY_UTF8MB4 "_sv_0900_as_cs", /* m_coll_name */

10709

"", /* comment */

10710

sv_cldr_30, /* tailoring */

10711

nullptr, /* coll_param */

10712

ctype_utf8, /* ctype */

10713

nullptr, /* to_lower */

10714

nullptr, /* to_upper */

10715

nullptr, /* sort_order */

10716

&my_uca_v900, /* uca */

10717

nullptr, /* tab_to_uni */

10718

nullptr, /* tab_from_uni */

10719

&my_unicase_unicode900, /* caseinfo */

10720

nullptr, /* state_map */

10721

nullptr, /* ident_map */

10722

0, /* strxfrm_multiply */

10723

1, /* caseup_multiply */

10724

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10729

0x10FFFF, /* max_sort_char */

10730

' ', /* pad char */

10731

false, /* escape_with_backslash_is_dangerous */

10732

3, /* levels_for_compare */

10733

&my_charset_utf8mb4_handler,

10734

&my_collation_uca_900_handler,

10735

NO_PAD};

10736

10737

CHARSET_INFO my_charset_utf8mb4_tr_0900_as_cs = {

288,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10742

MY_UTF8MB4, /* csname */

10743

MY_UTF8MB4 "_tr_0900_as_cs", /* m_coll_name */

10744

"", /* comment */

10745

tr_cldr_30, /* tailoring */

10746

nullptr, /* coll_param */

10747

ctype_utf8, /* ctype */

10748

nullptr, /* to_lower */

10749

nullptr, /* to_upper */

10750

nullptr, /* sort_order */

10751

&my_uca_v900, /* uca */

10752

nullptr, /* tab_to_uni */

10753

nullptr, /* tab_from_uni */

10754

&my_unicase_unicode900, /* caseinfo */

10755

nullptr, /* state_map */

10756

nullptr, /* ident_map */

10757

0, /* strxfrm_multiply */

10758

1, /* caseup_multiply */

10759

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10764

0x10FFFF, /* max_sort_char */

10765

' ', /* pad char */

10766

false, /* escape_with_backslash_is_dangerous */

10767

3, /* levels_for_compare */

10768

&my_charset_utf8mb4_handler,

10769

&my_collation_uca_900_handler,

10770

NO_PAD};

10771

10772

CHARSET_INFO my_charset_utf8mb4_cs_0900_as_cs = {

289,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10777

MY_UTF8MB4, /* csname */

10778

MY_UTF8MB4 "_cs_0900_as_cs", /* m_coll_name */

10779

"", /* comment */

10780

cs_cldr_30, /* tailoring */

10781

nullptr, /* coll_param */

10782

ctype_utf8, /* ctype */

10783

nullptr, /* to_lower */

10784

nullptr, /* to_upper */

10785

nullptr, /* sort_order */

10786

&my_uca_v900, /* uca */

10787

nullptr, /* tab_to_uni */

10788

nullptr, /* tab_from_uni */

10789

&my_unicase_unicode900, /* caseinfo */

10790

nullptr, /* state_map */

10791

nullptr, /* ident_map */

10792

0, /* strxfrm_multiply */

10793

1, /* caseup_multiply */

10794

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10799

0x10FFFF, /* max_sort_char */

10800

' ', /* pad char */

10801

false, /* escape_with_backslash_is_dangerous */

10802

3, /* levels_for_compare */

10803

&my_charset_utf8mb4_handler,

10804

&my_collation_uca_900_handler,

10805

NO_PAD};

10806

10807

CHARSET_INFO my_charset_utf8mb4_da_0900_as_cs = {

290,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10812

MY_UTF8MB4, /* csname */

10813

MY_UTF8MB4 "_da_0900_as_cs", /* m_coll_name */

10814

"", /* comment */

10815

da_cldr_30, /* tailoring */

10816

&da_coll_param, /* coll_param */

10817

ctype_utf8, /* ctype */

10818

nullptr, /* to_lower */

10819

nullptr, /* to_upper */

10820

nullptr, /* sort_order */

10821

&my_uca_v900, /* uca */

10822

nullptr, /* tab_to_uni */

10823

nullptr, /* tab_from_uni */

10824

&my_unicase_unicode900, /* caseinfo */

10825

nullptr, /* state_map */

10826

nullptr, /* ident_map */

10827

0, /* strxfrm_multiply */

10828

1, /* caseup_multiply */

10829

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10834

0x10FFFF, /* max_sort_char */

10835

' ', /* pad char */

10836

false, /* escape_with_backslash_is_dangerous */

10837

3, /* levels_for_compare */

10838

&my_charset_utf8mb4_handler,

10839

&my_collation_uca_900_handler,

10840

NO_PAD};

10841

10842

CHARSET_INFO my_charset_utf8mb4_lt_0900_as_cs = {

291,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10847

MY_UTF8MB4, /* csname */

10848

MY_UTF8MB4 "_lt_0900_as_cs", /* m_coll_name */

10849

"", /* comment */

10850

lt_cldr_30, /* tailoring */

10851

nullptr, /* coll_param */

10852

ctype_utf8, /* ctype */

10853

nullptr, /* to_lower */

10854

nullptr, /* to_upper */

10855

nullptr, /* sort_order */

10856

&my_uca_v900, /* uca */

10857

nullptr, /* tab_to_uni */

10858

nullptr, /* tab_from_uni */

10859

&my_unicase_unicode900, /* caseinfo */

10860

nullptr, /* state_map */

10861

nullptr, /* ident_map */

10862

0, /* strxfrm_multiply */

10863

1, /* caseup_multiply */

10864

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10869

0x10FFFF, /* max_sort_char */

10870

' ', /* pad char */

10871

false, /* escape_with_backslash_is_dangerous */

10872

3, /* levels_for_compare */

10873

&my_charset_utf8mb4_handler,

10874

&my_collation_uca_900_handler,

10875

NO_PAD};

10876

10877

CHARSET_INFO my_charset_utf8mb4_sk_0900_as_cs = {

292,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10882

MY_UTF8MB4, /* csname */

10883

MY_UTF8MB4 "_sk_0900_as_cs", /* m_coll_name */

10884

"", /* comment */

10885

sk_cldr_30, /* tailoring */

10886

nullptr, /* coll_param */

10887

ctype_utf8, /* ctype */

10888

nullptr, /* to_lower */

10889

nullptr, /* to_upper */

10890

nullptr, /* sort_order */

10891

&my_uca_v900, /* uca */

10892

nullptr, /* tab_to_uni */

10893

nullptr, /* tab_from_uni */

10894

&my_unicase_unicode900, /* caseinfo */

10895

nullptr, /* state_map */

10896

nullptr, /* ident_map */

10897

0, /* strxfrm_multiply */

10898

1, /* caseup_multiply */

10899

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10904

0x10FFFF, /* max_sort_char */

10905

' ', /* pad char */

10906

false, /* escape_with_backslash_is_dangerous */

10907

3, /* levels_for_compare */

10908

&my_charset_utf8mb4_handler,

10909

&my_collation_uca_900_handler,

10910

NO_PAD};

10911

10912

CHARSET_INFO my_charset_utf8mb4_es_trad_0900_as_cs = {

293,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10917

MY_UTF8MB4, /* csname */

10918

MY_UTF8MB4 "_es_trad_0900_as_cs", /* m_coll_name */

10919

"", /* comment */

10920

es_trad_cldr_30, /* tailoring */

10921

nullptr, /* coll_param */

10922

ctype_utf8, /* ctype */

10923

nullptr, /* to_lower */

10924

nullptr, /* to_upper */

10925

nullptr, /* sort_order */

10926

&my_uca_v900, /* uca */

10927

nullptr, /* tab_to_uni */

10928

nullptr, /* tab_from_uni */

10929

&my_unicase_unicode900, /* caseinfo */

10930

nullptr, /* state_map */

10931

nullptr, /* ident_map */

10932

0, /* strxfrm_multiply */

10933

1, /* caseup_multiply */

10934

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10939

0x10FFFF, /* max_sort_char */

10940

' ', /* pad char */

10941

false, /* escape_with_backslash_is_dangerous */

10942

3, /* levels_for_compare */

10943

&my_charset_utf8mb4_handler,

10944

&my_collation_uca_900_handler,

10945

NO_PAD};

10946

10947

CHARSET_INFO my_charset_utf8mb4_la_0900_as_cs = {

294,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

10952

MY_UTF8MB4, /* csname */

10953

MY_UTF8MB4 "_la_0900_as_cs", /* m_coll_name */

10954

"", /* comment */

10955

roman, /* tailoring */

10956

nullptr, /* coll_param */

10957

ctype_utf8, /* ctype */

10958

nullptr, /* to_lower */

10959

nullptr, /* to_upper */

10960

nullptr, /* sort_order */

10961

&my_uca_v900, /* uca */

10962

nullptr, /* tab_to_uni */

10963

nullptr, /* tab_from_uni */

10964

&my_unicase_unicode900, /* caseinfo */

10965

nullptr, /* state_map */

10966

nullptr, /* ident_map */

10967

0, /* strxfrm_multiply */

10968

1, /* caseup_multiply */

10969

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

10974

0x10FFFF, /* max_sort_char */

10975

' ', /* pad char */

10976

false, /* escape_with_backslash_is_dangerous */

10977

3, /* levels_for_compare */

10978

&my_charset_utf8mb4_handler,

10979

&my_collation_uca_900_handler,

NO_PAD};

#if 0

CHARSET_INFO my_charset_utf8mb4_fa_0900_as_cs=

10984

{

10985

295, 0, 0, /* number */

10986

MY_CS_UTF8MB4_UCA_FLAGS|MY_CS_CSSORT,/* state */

10987

MY_UTF8MB4, /* csname */

10988

MY_UTF8MB4 "_fa_0900_as_cs",/* m_coll_name */

10989

"", /* comment */

10990

fa_cldr_30, /* tailoring */

10991

&fa_coll_param, /* coll_param */

10992

ctype_utf8, /* ctype */

10993

NULL, /* to_lower */

10994

NULL, /* to_upper */

10995

NULL, /* sort_order */

10996

&my_uca_v900, /* uca */

10997

NULL, /* tab_to_uni */

10998

NULL, /* tab_from_uni */

10999

&my_unicase_unicode900,/* caseinfo */

11000

NULL, /* state_map */

11001

NULL, /* ident_map */

11002

0, /* strxfrm_multiply */

11003

1, /* caseup_multiply */

11004

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11009

0x10FFFF, /* max_sort_char */

11010

' ', /* pad char */

11011

0, /* escape_with_backslash_is_dangerous */

11012

3, /* levels_for_compare */

11013

&my_charset_utf8mb4_handler,

11014

&my_collation_uca_900_handler,

NO_PAD

};

#endif

CHARSET_INFO my_charset_utf8mb4_eo_0900_as_cs = {

296,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11024

MY_UTF8MB4, /* csname */

11025

MY_UTF8MB4 "_eo_0900_as_cs", /* m_coll_name */

11026

"", /* comment */

11027

esperanto, /* tailoring */

11028

nullptr, /* coll_param */

11029

ctype_utf8, /* ctype */

11030

nullptr, /* to_lower */

11031

nullptr, /* to_upper */

11032

nullptr, /* sort_order */

11033

&my_uca_v900, /* uca */

11034

nullptr, /* tab_to_uni */

11035

nullptr, /* tab_from_uni */

11036

&my_unicase_unicode900, /* caseinfo */

11037

nullptr, /* state_map */

11038

nullptr, /* ident_map */

11039

0, /* strxfrm_multiply */

11040

1, /* caseup_multiply */

11041

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11046

0x10FFFF, /* max_sort_char */

11047

' ', /* pad char */

11048

false, /* escape_with_backslash_is_dangerous */

11049

3, /* levels_for_compare */

11050

&my_charset_utf8mb4_handler,

11051

&my_collation_uca_900_handler,

11052

NO_PAD};

11053

11054

CHARSET_INFO my_charset_utf8mb4_hu_0900_as_cs = {

297,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11059

MY_UTF8MB4, /* csname */

11060

MY_UTF8MB4 "_hu_0900_as_cs", /* m_coll_name */

11061

"", /* comment */

11062

hu_cldr_30, /* tailoring */

11063

nullptr, /* coll_param */

11064

ctype_utf8, /* ctype */

11065

nullptr, /* to_lower */

11066

nullptr, /* to_upper */

11067

nullptr, /* sort_order */

11068

&my_uca_v900, /* uca */

11069

nullptr, /* tab_to_uni */

11070

nullptr, /* tab_from_uni */

11071

&my_unicase_unicode900, /* caseinfo */

11072

nullptr, /* state_map */

11073

nullptr, /* ident_map */

11074

0, /* strxfrm_multiply */

11075

1, /* caseup_multiply */

11076

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11081

0x10FFFF, /* max_sort_char */

11082

' ', /* pad char */

11083

false, /* escape_with_backslash_is_dangerous */

11084

3, /* levels_for_compare */

11085

&my_charset_utf8mb4_handler,

11086

&my_collation_uca_900_handler,

11087

NO_PAD};

11088

11089

CHARSET_INFO my_charset_utf8mb4_hr_0900_as_cs = {

298,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11094

MY_UTF8MB4, /* csname */

11095

MY_UTF8MB4 "_hr_0900_as_cs", /* m_coll_name */

11096

"", /* comment */

11097

hr_cldr_30, /* tailoring */

11098

&hr_coll_param, /* coll_param */

11099

ctype_utf8, /* ctype */

11100

nullptr, /* to_lower */

11101

nullptr, /* to_upper */

11102

nullptr, /* sort_order */

11103

&my_uca_v900, /* uca */

11104

nullptr, /* tab_to_uni */

11105

nullptr, /* tab_from_uni */

11106

&my_unicase_unicode900, /* caseinfo */

11107

nullptr, /* state_map */

11108

nullptr, /* ident_map */

11109

0, /* strxfrm_multiply */

11110

1, /* caseup_multiply */

11111

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11116

0x10FFFF, /* max_sort_char */

11117

' ', /* pad char */

11118

false, /* escape_with_backslash_is_dangerous */

11119

3, /* levels_for_compare */

11120

&my_charset_utf8mb4_handler,

11121

&my_collation_uca_900_handler,

NO_PAD};

#if 0

CHARSET_INFO my_charset_utf8mb4_si_0900_as_cs=

11126

{

11127

299, 0, 0, /* number */

11128

MY_CS_UTF8MB4_UCA_FLAGS|MY_CS_CSSORT,/* state */

11129

MY_UTF8MB4, /* csname */

11130

MY_UTF8MB4 "_si_0900_as_cs",/* m_coll_name */

11131

"", /* comment */

11132

si_cldr_30, /* tailoring */

11133

NULL, /* coll_param */

11134

ctype_utf8, /* ctype */

11135

NULL, /* to_lower */

11136

NULL, /* to_upper */

11137

NULL, /* sort_order */

11138

&my_uca_v900, /* uca */

11139

NULL, /* tab_to_uni */

11140

NULL, /* tab_from_uni */

11141

&my_unicase_unicode900,/* caseinfo */

11142

NULL, /* state_map */

11143

NULL, /* ident_map */

11144

0, /* strxfrm_multiply */

11145

1, /* caseup_multiply */

11146

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11151

0x10FFFF, /* max_sort_char */

11152

' ', /* pad char */

11153

0, /* escape_with_backslash_is_dangerous */

11154

3, /* levels_for_compare */

11155

&my_charset_utf8mb4_handler,

11156

&my_collation_uca_900_handler,

NO_PAD

};

#endif

CHARSET_INFO my_charset_utf8mb4_vi_0900_as_cs = {

300,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11166

MY_UTF8MB4, /* csname */

11167

MY_UTF8MB4 "_vi_0900_as_cs", /* m_coll_name */

11168

"", /* comment */

11169

vi_cldr_30, /* tailoring */

11170

&vi_coll_param, /* coll_param */

11171

ctype_utf8, /* ctype */

11172

nullptr, /* to_lower */

11173

nullptr, /* to_upper */

11174

nullptr, /* sort_order */

11175

&my_uca_v900, /* uca */

11176

nullptr, /* tab_to_uni */

11177

nullptr, /* tab_from_uni */

11178

&my_unicase_unicode900, /* caseinfo */

11179

nullptr, /* state_map */

11180

nullptr, /* ident_map */

11181

0, /* strxfrm_multiply */

11182

1, /* caseup_multiply */

11183

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11188

0x10FFFF, /* max_sort_char */

11189

' ', /* pad char */

11190

false, /* escape_with_backslash_is_dangerous */

11191

3, /* levels_for_compare */

11192

&my_charset_utf8mb4_handler,

11193

&my_collation_uca_900_handler,

11194

NO_PAD};

11195

11196

CHARSET_INFO my_charset_utf8mb4_ja_0900_as_cs = {

303,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11201

MY_UTF8MB4, /* csname */

11202

MY_UTF8MB4 "_ja_0900_as_cs", /* m_coll_name */

11203

"", /* comment */

11204

ja_cldr_30, /* tailoring */

11205

&ja_coll_param, /* coll_param */

11206

ctype_utf8, /* ctype */

11207

nullptr, /* to_lower */

11208

nullptr, /* to_upper */

11209

nullptr, /* sort_order */

11210

&my_uca_v900, /* uca */

11211

nullptr, /* tab_to_uni */

11212

nullptr, /* tab_from_uni */

11213

&my_unicase_unicode900, /* caseinfo */

11214

nullptr, /* state_map */

11215

nullptr, /* ident_map */

11216

0, /* strxfrm_multiply */

11217

1, /* caseup_multiply */

11218

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11223

0x10FFFF, /* max_sort_char */

11224

' ', /* pad char */

11225

false, /* escape_with_backslash_is_dangerous */

11226

3, /* levels_for_compare */

11227

&my_charset_utf8mb4_handler,

11228

&my_collation_uca_900_handler,

11229

NO_PAD};

11230

11231

CHARSET_INFO my_charset_utf8mb4_ja_0900_as_cs_ks = {

304,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11236

MY_UTF8MB4, /* csname */

11237

MY_UTF8MB4 "_ja_0900_as_cs_ks", /* m_coll_name */

11238

"", /* comment */

11239

ja_cldr_30, /* tailoring */

11240

&ja_coll_param, /* coll_param */

11241

ctype_utf8, /* ctype */

11242

nullptr, /* to_lower */

11243

nullptr, /* to_upper */

11244

nullptr, /* sort_order */

11245

&my_uca_v900, /* uca */

11246

nullptr, /* tab_to_uni */

11247

nullptr, /* tab_from_uni */

11248

&my_unicase_unicode900, /* caseinfo */

11249

nullptr, /* state_map */

11250

nullptr, /* ident_map */

11251

24, /* strxfrm_multiply */

11252

1, /* caseup_multiply */

11253

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11258

0x10FFFF, /* max_sort_char */

11259

' ', /* pad char */

11260

false, /* escape_with_backslash_is_dangerous */

11261

4, /* levels_for_compare */

11262

&my_charset_utf8mb4_handler,

11263

&my_collation_uca_900_handler,

11264

NO_PAD};

11265

11266

CHARSET_INFO my_charset_utf8mb4_0900_as_ci = {

305,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11271

MY_UTF8MB4, /* csname */

11272

MY_UTF8MB4 "_0900_as_ci", /* m_coll_name */

11273

"", /* comment */

11274

nullptr, /* tailoring */

11275

nullptr, /* coll_param */

11276

ctype_utf8, /* ctype */

11277

nullptr, /* to_lower */

11278

nullptr, /* to_upper */

11279

nullptr, /* sort_order */

11280

&my_uca_v900, /* uca */

11281

nullptr, /* tab_to_uni */

11282

nullptr, /* tab_from_uni */

11283

&my_unicase_unicode900, /* caseinfo */

11284

nullptr, /* state_map */

11285

nullptr, /* ident_map */

11286

0, /* strxfrm_multiply */

11287

1, /* caseup_multiply */

11288

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11293

0x10FFFF, /* max_sort_char */

11294

' ', /* pad char */

11295

false, /* escape_with_backslash_is_dangerous */

11296

2, /* levels_for_compare */

11297

&my_charset_utf8mb4_handler,

11298

&my_collation_uca_900_handler,

11299

NO_PAD};

11300

11301

CHARSET_INFO my_charset_utf8mb4_ru_0900_ai_ci = {

306,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11306

MY_UTF8MB4, /* csname */

11307

MY_UTF8MB4 "_ru_0900_ai_ci", /* m_coll_name */

11308

"", /* comment */

11309

"", /* tailoring */

11310

&ru_coll_param, /* coll_param */

11311

ctype_utf8, /* ctype */

11312

nullptr, /* to_lower */

11313

nullptr, /* to_upper */

11314

nullptr, /* sort_order */

11315

&my_uca_v900, /* uca */

11316

nullptr, /* tab_to_uni */

11317

nullptr, /* tab_from_uni */

11318

&my_unicase_unicode900, /* caseinfo */

11319

nullptr, /* state_map */

11320

nullptr, /* ident_map */

11321

0, /* strxfrm_multiply */

11322

1, /* caseup_multiply */

11323

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11328

0x10FFFF, /* max_sort_char */

11329

' ', /* pad char */

11330

false, /* escape_with_backslash_is_dangerous */

11331

1, /* levels_for_compare */

11332

&my_charset_utf8mb4_handler,

11333

&my_collation_uca_900_handler,

11334

NO_PAD};

11335

11336

CHARSET_INFO my_charset_utf8mb4_ru_0900_as_cs = {

307,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11341

MY_UTF8MB4, /* csname */

11342

MY_UTF8MB4 "_ru_0900_as_cs", /* m_coll_name */

11343

"", /* comment */

11344

"", /* tailoring */

11345

&ru_coll_param, /* coll_param */

11346

ctype_utf8, /* ctype */

11347

nullptr, /* to_lower */

11348

nullptr, /* to_upper */

11349

nullptr, /* sort_order */

11350

&my_uca_v900, /* uca */

11351

nullptr, /* tab_to_uni */

11352

nullptr, /* tab_from_uni */

11353

&my_unicase_unicode900, /* caseinfo */

11354

nullptr, /* state_map */

11355

nullptr, /* ident_map */

11356

0, /* strxfrm_multiply */

11357

1, /* caseup_multiply */

11358

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11363

0x10FFFF, /* max_sort_char */

11364

' ', /* pad char */

11365

false, /* escape_with_backslash_is_dangerous */

11366

3, /* levels_for_compare */

11367

&my_charset_utf8mb4_handler,

11368

&my_collation_uca_900_handler,

11369

NO_PAD};

11370

11371

CHARSET_INFO my_charset_utf8mb4_zh_0900_as_cs = {

308,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11376

MY_UTF8MB4, /* csname */

11377

MY_UTF8MB4 "_zh_0900_as_cs", /* m_coll_name */

11378

"", /* comment */

11379

zh_cldr_30, /* tailoring */

11380

&zh_coll_param, /* coll_param */

11381

ctype_utf8, /* ctype */

11382

nullptr, /* to_lower */

11383

nullptr, /* to_upper */

11384

nullptr, /* sort_order */

11385

&my_uca_v900, /* uca */

11386

nullptr, /* tab_to_uni */

11387

nullptr, /* tab_from_uni */

11388

&my_unicase_unicode900, /* caseinfo */

11389

nullptr, /* state_map */

11390

nullptr, /* ident_map */

11391

0, /* strxfrm_multiply */

11392

1, /* caseup_multiply */

11393

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11398

0x10FFFF, /* max_sort_char */

11399

' ', /* pad char */

11400

false, /* escape_with_backslash_is_dangerous */

11401

3, /* levels_for_compare */

11402

&my_charset_utf8mb4_handler,

11403

&my_collation_uca_900_handler,

NO_PAD};

/*

Comparing the UTF-8 representation automatically yields codepoint order,

11408

so we can just do a binary comparison. Note that

11409

my_strnxfrm_unicode_full_bin() chooses to transform to UCS before collation;

11410

this is purely for legacy reasons and is not needed here.

11411

*/

11412

3343047

static size_t my_strnxfrm_utf8mb4_0900_bin(const CHARSET_INFO *cs

11413

[[maybe_unused]],

11414

uchar *dst, size_t dstlen,

11415

uint nweights [[maybe_unused]],

11416

const uchar *src, size_t srclen,

11417

uint flags) {

11418

1/2

✗ Branch 0 not taken.

✓ Branch 1 taken 3343047 times.

3343047

assert(src);

11419

11420

3343047

size_t weight_len = std::min<size_t>(srclen, dstlen);

11421

3343047

memcpy(dst, src, weight_len);

11422

2/2

✓ Branch 0 taken 78 times.

✓ Branch 1 taken 3342969 times.

3343047

if (flags & MY_STRXFRM_PAD_TO_MAXLEN) {

11423

78

memset(dst + weight_len, 0, dstlen - weight_len);

11424

78

return dstlen;

11425

} else {

11426

3342969

return weight_len;

}

}

1377502

static int my_strnncollsp_utf8mb4_0900_bin(const CHARSET_INFO *cs,

11431

const uchar *s, size_t slen,

11432

const uchar *t, size_t tlen) {

11433

1377502

return my_strnncoll_mb_bin(cs, s, slen, t, tlen, false);

11434

}

11435

11436

static MY_COLLATION_HANDLER my_collation_utf8mb4_0900_bin_handler = {

nullptr, /* init */

nullptr,

my_strnncoll_mb_bin,

my_strnncollsp_utf8mb4_0900_bin,

11441

my_strnxfrm_utf8mb4_0900_bin,

11442

my_strnxfrmlen_simple,

11443

my_like_range_mb,

11444

my_wildcmp_mb_bin,

11445

my_strcasecmp_mb_bin,

11446

my_instr_mb,

11447

my_hash_sort_mb_bin,

11448

my_propagate_simple};

11449

11450

CHARSET_INFO my_charset_utf8mb4_0900_bin = {

309,

0,

0, // number

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_BINSORT, // state

11455

MY_UTF8MB4, // cs name

11456

MY_UTF8MB4 "_0900_bin", // name

11457

"", // comment

11458

nullptr, // tailoring

11459

nullptr, // coll_param

ctype_utf8, // ctype

nullptr, // to_lower

nullptr, // to_upper

nullptr, // sort_order

11464

nullptr, // uca

11465

nullptr, // tab_to_uni

11466

nullptr, // tab_from_uni

11467

&my_unicase_unicode900, // caseinfo

11468

nullptr, // state_map

11469

nullptr, // ident_map

11470

1, // strxfrm_multiply

11471

1, // caseup_multiply

11472

1, // casedn_multiply

1, // mbminlen

4, // mbmaxlen

1, // mbmaxlenlen

0, // min_sort_char

0x10FFFF, // max_sort_char

11478

' ', // pad char

11479

false, // escape_with_backslash_is_dangerous

11480

1, // levels_for_compare

11481

&my_charset_utf8mb4_handler,

11482

&my_collation_utf8mb4_0900_bin_handler,

11483

NO_PAD};

11484

11485

CHARSET_INFO my_charset_utf8mb4_nb_0900_ai_ci = {

310,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11490

MY_UTF8MB4, /* csname */

11491

MY_UTF8MB4 "_nb_0900_ai_ci", /* name */

11492

"", /* comment */

11493

da_cldr_30, /* tailoring */

11494

nullptr, /* coll_param */

11495

ctype_utf8, /* ctype */

11496

nullptr, /* to_lower */

11497

nullptr, /* to_upper */

11498

nullptr, /* sort_order */

11499

&my_uca_v900, /* uca */

11500

nullptr, /* tab_to_uni */

11501

nullptr, /* tab_from_uni */

11502

&my_unicase_unicode900, /* caseinfo */

11503

nullptr, /* state_map */

11504

nullptr, /* ident_map */

11505

0, /* strxfrm_multiply */

11506

1, /* caseup_multiply */

11507

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11512

0x10FFFF, /* max_sort_char */

11513

' ', /* pad char */

11514

false, /* escape_with_backslash_is_dangerous */

11515

1, /* levels_for_compare */

11516

&my_charset_utf8mb4_handler,

11517

&my_collation_uca_900_handler,

11518

NO_PAD};

11519

11520

CHARSET_INFO my_charset_utf8mb4_nb_0900_as_cs = {

311,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11525

MY_UTF8MB4, /* csname */

11526

MY_UTF8MB4 "_nb_0900_as_cs", /* name */

11527

"", /* comment */

11528

da_cldr_30, /* tailoring */

11529

&no_coll_param, /* coll_param */

11530

ctype_utf8, /* ctype */

11531

nullptr, /* to_lower */

11532

nullptr, /* to_upper */

11533

nullptr, /* sort_order */

11534

&my_uca_v900, /* uca */

11535

nullptr, /* tab_to_uni */

11536

nullptr, /* tab_from_uni */

11537

&my_unicase_unicode900, /* caseinfo */

11538

nullptr, /* state_map */

11539

nullptr, /* ident_map */

11540

0, /* strxfrm_multiply */

11541

1, /* caseup_multiply */

11542

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11547

0x10FFFF, /* max_sort_char */

11548

' ', /* pad char */

11549

false, /* escape_with_backslash_is_dangerous */

11550

3, /* levels_for_compare */

11551

&my_charset_utf8mb4_handler,

11552

&my_collation_uca_900_handler,

11553

NO_PAD};

11554

11555

CHARSET_INFO my_charset_utf8mb4_nn_0900_ai_ci = {

312,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11560

MY_UTF8MB4, /* csname */

11561

MY_UTF8MB4 "_nn_0900_ai_ci", /* name */

11562

"", /* comment */

11563

da_cldr_30, /* tailoring */

11564

nullptr, /* coll_param */

11565

ctype_utf8, /* ctype */

11566

nullptr, /* to_lower */

11567

nullptr, /* to_upper */

11568

nullptr, /* sort_order */

11569

&my_uca_v900, /* uca */

11570

nullptr, /* tab_to_uni */

11571

nullptr, /* tab_from_uni */

11572

&my_unicase_unicode900, /* caseinfo */

11573

nullptr, /* state_map */

11574

nullptr, /* ident_map */

11575

0, /* strxfrm_multiply */

11576

1, /* caseup_multiply */

11577

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11582

0x10FFFF, /* max_sort_char */

11583

' ', /* pad char */

11584

false, /* escape_with_backslash_is_dangerous */

11585

1, /* levels_for_compare */

11586

&my_charset_utf8mb4_handler,

11587

&my_collation_uca_900_handler,

11588

NO_PAD};

11589

11590

CHARSET_INFO my_charset_utf8mb4_nn_0900_as_cs = {

313,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11595

MY_UTF8MB4, /* csname */

11596

MY_UTF8MB4 "_nn_0900_as_cs", /* name */

11597

"", /* comment */

11598

da_cldr_30, /* tailoring */

11599

&no_coll_param, /* coll_param */

11600

ctype_utf8, /* ctype */

11601

nullptr, /* to_lower */

11602

nullptr, /* to_upper */

11603

nullptr, /* sort_order */

11604

&my_uca_v900, /* uca */

11605

nullptr, /* tab_to_uni */

11606

nullptr, /* tab_from_uni */

11607

&my_unicase_unicode900, /* caseinfo */

11608

nullptr, /* state_map */

11609

nullptr, /* ident_map */

11610

0, /* strxfrm_multiply */

11611

1, /* caseup_multiply */

11612

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11617

0x10FFFF, /* max_sort_char */

11618

' ', /* pad char */

11619

false, /* escape_with_backslash_is_dangerous */

11620

3, /* levels_for_compare */

11621

&my_charset_utf8mb4_handler,

11622

&my_collation_uca_900_handler,

11623

NO_PAD};

11624

11625

CHARSET_INFO my_charset_utf8mb4_sr_latn_0900_ai_ci = {

314,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11630

MY_UTF8MB4, /* csname */

11631

MY_UTF8MB4 "_sr_latn_0900_ai_ci", /* name */

11632

"", /* comment */

11633

hr_cldr_30, /* tailoring */

11634

&hr_coll_param, /* coll_param */

11635

ctype_utf8, /* ctype */

11636

nullptr, /* to_lower */

11637

nullptr, /* to_upper */

11638

nullptr, /* sort_order */

11639

&my_uca_v900, /* uca */

11640

nullptr, /* tab_to_uni */

11641

nullptr, /* tab_from_uni */

11642

&my_unicase_unicode900, /* caseinfo */

11643

nullptr, /* state_map */

11644

nullptr, /* ident_map */

11645

0, /* strxfrm_multiply */

11646

1, /* caseup_multiply */

11647

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11652

0x10FFFF, /* max_sort_char */

11653

' ', /* pad char */

11654

false, /* escape_with_backslash_is_dangerous */

11655

1, /* levels_for_compare */

11656

&my_charset_utf8mb4_handler,

11657

&my_collation_uca_900_handler,

11658

NO_PAD};

11659

11660

CHARSET_INFO my_charset_utf8mb4_sr_latn_0900_as_cs = {

315,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11665

MY_UTF8MB4, /* csname */

11666

MY_UTF8MB4 "_sr_latn_0900_as_cs", /* name */

11667

"", /* comment */

11668

hr_cldr_30, /* tailoring */

11669

&hr_coll_param, /* coll_param */

11670

ctype_utf8, /* ctype */

11671

nullptr, /* to_lower */

11672

nullptr, /* to_upper */

11673

nullptr, /* sort_order */

11674

&my_uca_v900, /* uca */

11675

nullptr, /* tab_to_uni */

11676

nullptr, /* tab_from_uni */

11677

&my_unicase_unicode900, /* caseinfo */

11678

nullptr, /* state_map */

11679

nullptr, /* ident_map */

11680

0, /* strxfrm_multiply */

11681

1, /* caseup_multiply */

11682

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11687

0x10FFFF, /* max_sort_char */

11688

' ', /* pad char */

11689

false, /* escape_with_backslash_is_dangerous */

11690

3, /* levels_for_compare */

11691

&my_charset_utf8mb4_handler,

11692

&my_collation_uca_900_handler,

11693

NO_PAD};

11694

11695

CHARSET_INFO my_charset_utf8mb4_bs_0900_ai_ci = {

316,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11700

MY_UTF8MB4, /* csname */

11701

MY_UTF8MB4 "_bs_0900_ai_ci", /* name */

11702

"", /* comment */

11703

hr_cldr_30, /* tailoring */

11704

&hr_coll_param, /* coll_param */

11705

ctype_utf8, /* ctype */

11706

nullptr, /* to_lower */

11707

nullptr, /* to_upper */

11708

nullptr, /* sort_order */

11709

&my_uca_v900, /* uca */

11710

nullptr, /* tab_to_uni */

11711

nullptr, /* tab_from_uni */

11712

&my_unicase_unicode900, /* caseinfo */

11713

nullptr, /* state_map */

11714

nullptr, /* ident_map */

11715

0, /* strxfrm_multiply */

11716

1, /* caseup_multiply */

11717

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11722

0x10FFFF, /* max_sort_char */

11723

' ', /* pad char */

11724

false, /* escape_with_backslash_is_dangerous */

11725

1, /* levels_for_compare */

11726

&my_charset_utf8mb4_handler,

11727

&my_collation_uca_900_handler,

11728

NO_PAD};

11729

11730

CHARSET_INFO my_charset_utf8mb4_bs_0900_as_cs = {

317,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11735

MY_UTF8MB4, /* csname */

11736

MY_UTF8MB4 "_bs_0900_as_cs", /* name */

11737

"", /* comment */

11738

hr_cldr_30, /* tailoring */

11739

&hr_coll_param, /* coll_param */

11740

ctype_utf8, /* ctype */

11741

nullptr, /* to_lower */

11742

nullptr, /* to_upper */

11743

nullptr, /* sort_order */

11744

&my_uca_v900, /* uca */

11745

nullptr, /* tab_to_uni */

11746

nullptr, /* tab_from_uni */

11747

&my_unicase_unicode900, /* caseinfo */

11748

nullptr, /* state_map */

11749

nullptr, /* ident_map */

11750

0, /* strxfrm_multiply */

11751

1, /* caseup_multiply */

11752

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11757

0x10FFFF, /* max_sort_char */

11758

' ', /* pad char */

11759

false, /* escape_with_backslash_is_dangerous */

11760

3, /* levels_for_compare */

11761

&my_charset_utf8mb4_handler,

11762

&my_collation_uca_900_handler,

11763

NO_PAD};

11764

11765

CHARSET_INFO my_charset_utf8mb4_bg_0900_ai_ci = {

318,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11770

MY_UTF8MB4, /* csname */

11771

MY_UTF8MB4 "_bg_0900_ai_ci", /* name */

11772

"", /* comment */

11773

"", /* tailoring */

11774

&ru_coll_param, /* coll_param */

11775

ctype_utf8, /* ctype */

11776

nullptr, /* to_lower */

11777

nullptr, /* to_upper */

11778

nullptr, /* sort_order */

11779

&my_uca_v900, /* uca */

11780

nullptr, /* tab_to_uni */

11781

nullptr, /* tab_from_uni */

11782

&my_unicase_unicode900, /* caseinfo */

11783

nullptr, /* state_map */

11784

nullptr, /* ident_map */

11785

0, /* strxfrm_multiply */

11786

1, /* caseup_multiply */

11787

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11792

0x10FFFF, /* max_sort_char */

11793

' ', /* pad char */

11794

false, /* escape_with_backslash_is_dangerous */

11795

1, /* levels_for_compare */

11796

&my_charset_utf8mb4_handler,

11797

&my_collation_uca_900_handler,

11798

NO_PAD};

11799

11800

CHARSET_INFO my_charset_utf8mb4_bg_0900_as_cs = {

319,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11805

MY_UTF8MB4, /* csname */

11806

MY_UTF8MB4 "_bg_0900_as_cs", /* name */

11807

"", /* comment */

11808

"", /* tailoring */

11809

&ru_coll_param, /* coll_param */

11810

ctype_utf8, /* ctype */

11811

nullptr, /* to_lower */

11812

nullptr, /* to_upper */

11813

nullptr, /* sort_order */

11814

&my_uca_v900, /* uca */

11815

nullptr, /* tab_to_uni */

11816

nullptr, /* tab_from_uni */

11817

&my_unicase_unicode900, /* caseinfo */

11818

nullptr, /* state_map */

11819

nullptr, /* ident_map */

11820

0, /* strxfrm_multiply */

11821

1, /* caseup_multiply */

11822

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11827

0x10FFFF, /* max_sort_char */

11828

' ', /* pad char */

11829

false, /* escape_with_backslash_is_dangerous */

11830

3, /* levels_for_compare */

11831

&my_charset_utf8mb4_handler,

11832

&my_collation_uca_900_handler,

11833

NO_PAD};

11834

11835

CHARSET_INFO my_charset_utf8mb4_gl_0900_ai_ci = {

320,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11840

MY_UTF8MB4, /* csname */

11841

MY_UTF8MB4 "_gl_0900_ai_ci", /* name */

11842

"", /* comment */

11843

spanish, /* tailoring */

11844

nullptr, /* coll_param */

11845

ctype_utf8, /* ctype */

11846

nullptr, /* to_lower */

11847

nullptr, /* to_upper */

11848

nullptr, /* sort_order */

11849

&my_uca_v900, /* uca */

11850

nullptr, /* tab_to_uni */

11851

nullptr, /* tab_from_uni */

11852

&my_unicase_unicode900, /* caseinfo */

11853

nullptr, /* state_map */

11854

nullptr, /* ident_map */

11855

0, /* strxfrm_multiply */

11856

1, /* caseup_multiply */

11857

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11862

0x10FFFF, /* max_sort_char */

11863

' ', /* pad char */

11864

false, /* escape_with_backslash_is_dangerous */

11865

1, /* levels_for_compare */

11866

&my_charset_utf8mb4_handler,

11867

&my_collation_uca_900_handler,

11868

NO_PAD};

11869

11870

CHARSET_INFO my_charset_utf8mb4_gl_0900_as_cs = {

321,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11875

MY_UTF8MB4, /* csname */

11876

MY_UTF8MB4 "_gl_0900_as_cs", /* name */

11877

"", /* comment */

11878

spanish, /* tailoring */

11879

nullptr, /* coll_param */

11880

ctype_utf8, /* ctype */

11881

nullptr, /* to_lower */

11882

nullptr, /* to_upper */

11883

nullptr, /* sort_order */

11884

&my_uca_v900, /* uca */

11885

nullptr, /* tab_to_uni */

11886

nullptr, /* tab_from_uni */

11887

&my_unicase_unicode900, /* caseinfo */

11888

nullptr, /* state_map */

11889

nullptr, /* ident_map */

11890

0, /* strxfrm_multiply */

11891

1, /* caseup_multiply */

11892

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

9, /* min_sort_char */

11897

0x10FFFF, /* max_sort_char */

11898

' ', /* pad char */

11899

false, /* escape_with_backslash_is_dangerous */

11900

3, /* levels_for_compare */

11901

&my_charset_utf8mb4_handler,

11902

&my_collation_uca_900_handler,

11903

NO_PAD};

11904

11905

CHARSET_INFO my_charset_utf8mb4_mn_cyrl_0900_ai_ci = {

322,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS, /* state */

11910

MY_UTF8MB4, /* csname */

11911

MY_UTF8MB4 "_mn_cyrl_0900_ai_ci", /* name */

11912

"", /* comment */

11913

"", /* tailoring */

11914

&ru_coll_param, /* coll_param */

11915

ctype_utf8, /* ctype */

11916

nullptr, /* to_lower */

11917

nullptr, /* to_upper */

11918

nullptr, /* sort_order */

11919

&my_uca_v900, /* uca */

11920

nullptr, /* tab_to_uni */

11921

nullptr, /* tab_from_uni */

11922

&my_unicase_unicode900, /* caseinfo */

11923

nullptr, /* state_map */

11924

nullptr, /* ident_map */

11925

0, /* strxfrm_multiply */

11926

1, /* caseup_multiply */

11927

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11932

0x10FFFF, /* max_sort_char */

11933

' ', /* pad char */

11934

false, /* escape_with_backslash_is_dangerous */

11935

1, /* levels_for_compare */

11936

&my_charset_utf8mb4_handler,

11937

&my_collation_uca_900_handler,

11938

NO_PAD};

11939

11940

CHARSET_INFO my_charset_utf8mb4_mn_cyrl_0900_as_cs = {

323,

0,

0, /* number */

MY_CS_UTF8MB4_UCA_FLAGS | MY_CS_CSSORT, /* state */

11945

MY_UTF8MB4, /* csname */

11946

MY_UTF8MB4 "_mn_cyrl_0900_as_cs", /* name */

11947

"", /* comment */

11948

"", /* tailoring */

11949

&ru_coll_param, /* coll_param */

11950

ctype_utf8, /* ctype */

11951

nullptr, /* to_lower */

11952

nullptr, /* to_upper */

11953

nullptr, /* sort_order */

11954

&my_uca_v900, /* uca */

11955

nullptr, /* tab_to_uni */

11956

nullptr, /* tab_from_uni */

11957

&my_unicase_unicode900, /* caseinfo */

11958

nullptr, /* state_map */

11959

nullptr, /* ident_map */

11960

0, /* strxfrm_multiply */

11961

1, /* caseup_multiply */

11962

1, /* casedn_multiply */

1, /* mbminlen */

4, /* mbmaxlen */

1, /* mbmaxlenlen */

32, /* min_sort_char */

11967

0x10FFFF, /* max_sort_char */

11968

' ', /* pad char */

11969

false, /* escape_with_backslash_is_dangerous */

11970

3, /* levels_for_compare */

11971

&my_charset_utf8mb4_handler,

11972

&my_collation_uca_900_handler,

11973

NO_PAD};

11974